Push Pull Legs, Rebuilding After Three Weeks Off
Switching from seventeen years of a bro split to push pull legs after three weeks off training, documented weekly as the block runs.
Introduction
I have trained the same way since I was eighteen. Chest Monday, back Tuesday, legs Wednesday, shoulders Thursday, arms Friday, legs or chest Saturday. Seventeen years of the same structure, always training by feel rather than to a written programme. It worked. I got strong on it, hit lifetime personal records last year, and documented the numbers in Strength Snapshot, January 2026.
Then I went to Bodrum and stopped training for three weeks.
I walked around 20,000 steps a day while I was there, swam most mornings, and did nothing resembling resistance training. When I came back my bench had dropped from 100kg for 8 to 10 reps down to 3. People who see me regularly mentioned I had lost size. The first week back on the bro split confirmed it. Everything felt heavy and nothing moved the way it used to.
That layoff is the reason this post exists. Coming back to a blank slate is the cheapest opportunity I will get to change how I train, because there is nothing left to preserve. The numbers are already down.
So I am switching to push pull legs, running six sessions a week, and documenting the whole thing as I go.
Scope
This is not a programme review or a transformation post. It is a record of learning a different training structure after seventeen years of the same one. I am logging what I do, what changes, and what surprises me. The six primary lifts get recorded weekly so the option to analyse them properly stays open, however the point of this post is the mechanics of the switch rather than proving one structure beats another.
What the Layoff Cost
Three weeks off is not a deload. A deload reduces training stress while keeping the pattern alive. This was a full stop, and the 20,000 daily steps meant I was not resting either. I swapped one stressor for another.
| Metric | March 2026 | Late July 2026 |
|---|---|---|
| Bench press | 100kg × 8 to 10 | 100kg × 3 |
| Bodyweight | 79.7kg | 77.3kg |
Most of what I lost is neural rather than muscular. Strength expression drops fast when you stop training because the nervous system stops practising the movement. It also returns fast, usually inside three to four weeks. The tissue underneath adapts on a slower timeline, which is why the first week back is the wrong time to find out how strong I still am.
Why Push Pull Legs
Three reasons, in order of how much they mattered.
Frequency. My bench improved most during periods when I hit it twice a week. That is the clearest signal in my own training data. A bro split gives every muscle group one session per week. Push pull legs across six days gives each pattern two.
The split was carrying dead weight. A dedicated arms day on Friday was never earning its slot. Arms already get worked on every pressing and pulling session. Removing the day frees a session for a pattern that actually needs the frequency.
I have never trained this way. Seventeen years of the same structure is a long time to never test an alternative. The layoff removed the only real argument against switching, which was not wanting to interrupt progress.
Push, pull, legs
Push covers what the chest, shoulders and triceps do when the arms move away from the body. Bench, overhead press, dips, lateral raises, triceps work. Pull covers what the back and biceps do when the arms move toward the body. Rows, pull ups, face pulls, curls. Legs covers squats, deadlifts and everything below the waist. The split groups movements that already share muscles, so a session never trains a muscle that is still sore from the day before.
Structure
Each pattern runs twice a week, with the two sessions doing different jobs.
| Day | Session | Focus |
|---|---|---|
| Monday | Push A | Heavy, bench leads |
| Tuesday | Pull A | Heavy, row leads |
| Wednesday | Legs A | Heavy, squat leads |
| Thursday | Push B | Volume, shoulder press leads |
| Friday | Pull B | Volume, pull up leads |
| Saturday | Legs B | Volume, deadlift leads |
| Sunday | Rest |
The A sessions carry one top set in a low rep range followed by backoff work. The B sessions run higher reps at lighter loads with a different lead lift. Splitting the week this way means each pattern gets one session that builds strength and one that builds volume, rather than two identical sessions competing for the same recovery.
Only the top set goes near failure. Everything else stops two to three reps short. The intent is to accumulate work without accumulating fatigue that carries into the next day, which matters more when the same pattern comes around again in 72 hours instead of a week.
Double progression
Every exercise has a rep range rather than a fixed number. Bench backoff sets are 6 to 8 reps. I add reps at the same weight until every set hits 8, then add 2.5 to 5kg and drop back to 6. The weight only moves when the reps are earned. This removes the guesswork of when to add load and stops progression being a decision made under fatigue.
Heavy First, Then Volume
The biggest mechanical change from how I have always trained is set order. My habit was to pyramid up. Sets of 8 at a moderate weight, then heavier, then heavier again, stopping when the weight told me to stop.
Under a programme the top set is a number I am trying to move week to week, so it has to be taken under consistent conditions. Taking it last means measuring it under whatever fatigue the previous four sets left behind. On my first legs session I squatted 100kg for 8 across four sets, then loaded 120kg and got 4. Done in the other order, 120kg would almost certainly have given me more, and that difference is set order rather than strength.
Backoff sets do not have the same problem. They are submaximal by design, so a bit of accumulated fatigue is irrelevant to what they are for.
What counts as a warm up
A warm up set is one you finish and feel nothing from. The mechanical rule is reps. Warm ups stay at 1 to 5 reps even when the weight is light, because that is what keeps them free. 100kg for 8 is a working set no matter how light 100kg feels. 100kg for 2 is a warm up. Rest is the other tell. Warm ups need 30 to 60 seconds between them, working sets need two to three minutes.
What Kind of Programme This Is
This is a hypertrophy biased strength programme. Both, weighted toward size.
The strength side is the top set. One heavy set in a low rep range on a fixed lift, taken fresh, progressed week to week. That is how strength is built and how it gets tracked.
The hypertrophy side is everything after it. Four backoff sets plus five or six accessories at 8 to 15 reps is a lot of volume, and volume in that range is what drives size. A pure strength programme would be three or four exercises, heavy, and done. This has eight.
Hypertrophy and strength
Hypertrophy is growth in muscle tissue, meaning more fibre and more cross sectional area. Strength is the ability to express force with the tissue you already have, driven by motor unit recruitment, firing rate and coordination. They are related without being the same thing. A bigger muscle has a higher strength ceiling, however you have to train specifically to reach it.
The distinction explains what the layoff actually cost me. My bench fell from 100kg for 8 to 10 reps down to 3 in three weeks. Muscle does not disappear that fast. What disappeared was the ability to use it, because strength is quick to gain and quick to lose while tissue is slow on both counts.
| Hypertrophy bias | Strength bias | |
|---|---|---|
| Reps | 6 to 15 | 1 to 5 |
| Volume | High | Lower |
| Recovery cost | Moderate, spread out | High, concentrated |
| What improves | Size, work capacity, joint tolerance | Max lifts, expression of existing size |
| Timeline | Months | Weeks |
The bias makes sense for where I am. Seventeen years in, technique is not the limiting factor anymore. Tissue is. Adding size raises the ceiling that a strength block can then express. Hypertrophy work also recovers better across six sessions a week, because sets of 8 at 80 percent accumulate work without the neural cost of grinding heavy singles six days running.
So the sequencing is deliberate. Run this block hypertrophy biased and rebuild the tissue past where it was. Run a strength block after it with lower reps and fewer exercises to express what got built. That second block is its own post.
Rest and Recovery
Rest is three separate problems with three different answers, and the mistake is treating them as one.
Between sets. Full recovery is not the goal everywhere. The top set needs it, because a top set is a measurement and a measurement taken tired measures fatigue instead of strength. Most of what is available arrives inside three minutes and the curve flattens after that, so three to four minutes is the whole of it. Backoff and accessory work is the opposite case. Incomplete recovery is part of what makes those sets work, and stretching the rest until every set holds the same reps removes the reason they are in the session. My habit for seventeen years was to start a set when I felt ready, which quietly turned every exercise into a top set.
Between sessions. Muscle protein synthesis stays elevated for roughly 24 to 48 hours after a session. Push pull legs brings the same pattern back around every 72, which clears that window with room. The slower system is neural rather than muscular, and heavy work close to failure is what taxes it. That is the reason only one set per session goes near failure. Eight exercises of hard but submaximal work costs far less than three exercises of grinding singles.
Between weeks. One full day, Sunday, plus whatever sleep is worth. Sleep is the recovery variable that nothing in a programme compensates for. Connective tissue is the other consideration and it adapts on a slower clock than muscle does. Tendons and ligaments lag behind the strength that loads them, which is why the risk window on a block like this sits around weeks four to six rather than week one, and why the loads went up in small steps rather than back to March numbers on day one.
Rest is a variable, not a comfort
Resting four minutes one week and two the next moves the reps on the top set for reasons that have nothing to do with strength. If the number is being tracked, the conditions around it have to be fixed. A timer is the cheapest way to make a training log mean something.
Creatine
I am also back on creatine, 20g a day for seven days to load, then 5g daily. I ran a twelve week creatine study earlier this year, so this is a return to maintenance rather than a new variable to test.
It does mean two things changed at once. The training structure and the supplementation both started in the same week, so nothing in this log can be cleanly attributed to either. Anything that improves over the next fortnight is detraining reversal rather than creatine, based on what the loading phase looked like last time. The plan is to stay on 5g maintenance through whatever I run after this block so it stops being a variable at all.
Prediction, written before the first session
I think hitting each muscle group twice a week plus creatine makes me stronger than the bro split ever did. Seventeen years of training how I feel rather than to a programme, and I still got strong on it. Adding actual structure to that, with diet, sleep and no alcohol already in place, should make the difference. Writing this now so it stays honest either way.
The False Start
The plan was one week at reduced intensity before the block started properly. Everything capped three reps short of failure, six sessions, no maxes. The point was to give connective tissue a week to catch up with a nervous system that recovers faster than it does.
I got two sessions in.
Monday, Legs
| Exercise | Performance |
|---|---|
| Squat | 100kg × 8 for 4 sets, then 120kg × 4 |
| Romanian deadlift | 60kg × 8 for 3 sets |
| Leg press | 120kg × 12 for 3 sets |
| Nordic curl | Bodyweight × 8 for 3 sets |
| Calf raise | 50kg × 12 for 4 sets |
Romanian deadlifts were harder than expected. Not a strength problem, a coordination one. I have deadlifted from the floor for years and my hamstrings are not used to loading in that pattern.
The session felt light. I said so at the time, and it is worth having written down before I knew how the block would go.
Then the week ended early. A friend’s vaccinated dog bit the inside of my left wrist, breaking the skin. Over the next 48 hours the area became increasingly red, swollen and painful, with the redness spreading up my forearm. A personal trainer told me to get it checked. I went to hospital and was diagnosed with an infected dog bite.
Three days of IV antibiotics followed, the wound was cleaned and dressed repeatedly, and the plastics team monitored it. Inflammatory markers improved each day, the redness receded, pain and wrist movement improved, and I was discharged on a five day oral course with no surgery required.
That took out the rest of the week. I kept loading creatine throughout and ate more than usual while not training.
Neither of those two sessions counts toward the block. They were capped short of failure, cut off after two days, and the numbers came from a week I was about to spend in hospital. The block starts from the following Monday.
Week 1
Back in with minor soreness in the wrist and the infection cleared. Bodyweight 79kg on Monday morning, up 1.7kg from 77.3kg the week before. That is creatine water and glycogen rather than tissue, which is the same pattern the loading phase produced last time.
I also raised the volume before starting. The original layout ran six exercises per session, which is defensible on paper because each pattern now gets trained twice. However it sat well below what I am calibrated to after seventeen years, and dropping that far that fast is its own kind of shock. Each session now runs eight exercises.
Monday, Push A
| Exercise | Performance |
|---|---|
| Bench press top set | 100kg × 3 |
| Bench press backoff | 85kg × 8, 8, 6, 5 |
| Standing shoulder press | 50kg × 8, 6, 5 |
| Incline dumbbell press | 32kg × 8, 8, 5 |
| Chest fly | 66kg × 12 for 3 sets |
| Cable lateral raise | 20kg × 15 |
| Triceps pushdown | 80kg × 12 for 3 sets |
| Dips | Skipped, wrist |
Every compound dropped below its target rep range by the last set. That is a fatigue signature rather than a strength problem, and it is what a week off plus a first proper session at eight exercises produces. The isolation work held its reps because it is less systemically demanding.
The practical read is that 85kg for four sets of 6 to 8 is currently at the edge of what I can hold. No weight increase next Push A. The work is getting all four sets to 6, then 7, then 8, and only then moving to 90kg.
Tuesday, Pull A
| Exercise | Performance |
|---|---|
| Bent over row top set | 100kg × 6 |
| Bent over row backoff | 80kg × 8 for 4 sets |
| Pull ups to failure | 15, taken after rows |
| Pull ups backoff | 8 for 3 sets |
| Chest supported row | 20kg × 12 for 3 sets |
| Shrug | 34kg × 15 for 3 sets |
| Lat pulldown | 45kg × 15, 52kg × 15, 59kg × 12 |
| Face pull | 25kg × 15 for 3 sets |
| Hammer curl | 12kg × 12 for 3 sets |
Better than Push looked. The row hit the top of its range on day one and the backoffs held all four sets at 8 without dropping, which means 80kg is too light. Next Pull A goes to a 105kg top set and 85kg backoffs.
15 pull ups after five sets of heavy rows is a good number. I was at 20 to 21 fresh in March, so under fatigue and after three weeks off most of it is still there. Back held up considerably better through the layoff than pressing did, which is the first real finding of the block.
The lat pulldown is the old habit showing through. I ramped 45, 52, 59 with reps falling, which is pyramiding rather than the three sets at one weight the programme calls for. Corrected from next week.
I have no dip belt, so bodyweight pull ups to failure are the tracked top set for now. That number is already in my archive from January and March, which makes it useful. It moves to weighted once reps climb back above 15 taken fresh.
Wednesday, Legs A
| Exercise | Performance |
|---|---|
| Squat top set | 130kg × 4 |
| Squat backoff | 110kg × 8 for 4 sets |
| Front squat | 60kg × 8 for 4 sets |
| Romanian deadlift | 70kg × 10 for 4 sets |
| Leg press | 140kg × 12 for 4 sets |
| Walking lunge | 20kg × 12 for 4 sets |
| Nordic curl | Bodyweight × 8 for 4 sets |
| Calf raise | 50kg × 15 for 3 sets |
This is the session that proved the set order point. Last week I squatted 100kg for 8 across four sets and then loaded 120kg for 4. This week the heavy set came first and gave me 130kg for 4. Same reps, 10kg more, and nothing about my legs changed in eight days.
Backoffs held 8 reps across all four sets at 110kg, which means the weight is too light. It goes to 120kg next time. Calf raises the same.
Thursday, Push B
| Exercise | Performance |
|---|---|
| Standing shoulder press top set | 60kg × 6 |
| Bench press | 80kg × 12, 10, 7, 5 |
| Incline dumbbell press | 28kg × 12, 9, 8 |
| Seated dumbbell shoulder press | 20kg × 12, 9, 10 |
| Machine chest press | 80kg × 12 for 3 sets |
| Cable lateral raise | 9kg × 15 for 3 sets |
| Overhead triceps extension | 32kg × 12 for 3 sets |
60kg for 6 on the standing press is the best number of the week. I was at 70kg × 3 in January, so pressing overhead is sitting closer to baseline than flat bench is.
Bench went 12, 10, 7, 5. That is the third session running where pressing volume collapses across sets while everything else holds, which makes it a pattern rather than a bad day. The correction is to stop chasing the top of the range on set one. Opening at 12 and finishing on 5 is 34 reps of falling quality. Four solid sets of 8 is 32 reps that all count, and it progresses cleanly from there.
I also dropped press ups to failure from the programme. I never liked high rep work taken to failure and I think I only did it in the last plan for ego. It costs me the comparison against 80 reps in January, which I can get back cheaply by testing them once at the end of the block rather than every week.
Friday, Pull B
| Exercise | Performance |
|---|---|
| Pull ups top set | 22 to failure, taken fresh |
| Bent over row | 70kg × 12 for 4 sets |
| Single arm dumbbell row | 40kg × 12, 12, 10 |
| Lat pulldown | 52kg × 12 for 3 sets |
| Chest supported row | 25kg × 15 for 3 sets |
| Shrug | 40kg × 15 for 3 sets |
| Face pull | 50kg × 15 for 3 sets |
| Hammer curl | 14kg × 12 for 3 sets |
22 pull ups is a lifetime best. I was at 20 in January and 21 at the end of the creatine study in March, and this came after three weeks of not training at all.
The reason is set order again, not creatine and not the programme. Tuesday’s 15 came after five sets of heavy rows. Friday’s 22 came first in the session with two easy warm up sets in front of it. That is a 7 rep swing on the same lift three days apart, caused entirely by where it sat in the session. Two clean examples of the same principle inside one week.
The other read is that my back came through the layoff intact. Pull ups above lifetime best while bench sits at a third of what it was is a large divergence between two patterns that stopped training on the same day.
Rows at 70kg held 12 across all four sets, so that is too light as well. It goes to 80kg next Pull B.
The session started easy and got harder toward the end, which is the design working. Eight exercises with short rest on the back half is meant to accumulate.
Saturday, Legs B
| Exercise | Performance |
|---|---|
| Deadlift top set | 160kg × 3 |
| Squat | 100kg × 10 for 3 sets |
| Squat machine | 40kg × 10 for 3 sets |
| Leg curl | 32kg × 15 for 3 sets |
| Hip thrust | 40kg × 12 for 3 sets |
| Back extension | Bodyweight × 12 for 3 sets |
| Calf raise | 50kg × 20 for 4 sets |
160kg for 3 fills the last gap in the week. It is also deliberately low. I pulled 200kg in March and the bar had not been off the floor in over a month, so the top set was written at a number I could finish rather than a number I wanted. Deadlift carries the highest cost of getting it wrong of anything in the programme. Log it, go up next week.
Two exercises changed on the floor. Bulgarian split squats came out for a squat machine at 40kg, and the hip thrust went on the booty builder rather than a bar across the hips. The hip thrust swap is close enough to like for like. The other one is not. A Bulgarian split squat loads one leg at a time and makes balance part of the work, and a machine removes both of those. That was the only unilateral exercise in the week, so it goes back in next Legs B.
Three exercises finished at the ceiling of their rep range. Squats held 10 across all three sets, calf raises held 20 across four, leg curls held 15 across three. All three go up.
The session was comfortable the whole way through, which is the difference between it and Friday. Friday started easy and got harder because eight exercises on short rest are supposed to accumulate. Saturday never did. So that is the loads being light rather than the design being wrong, and it is the third session in the same week to say the same thing.
Week 2
Bodyweight 78.7kg on Monday morning, down 0.3kg on the week. The creatine load finished at the end of Week 1, so that is where the water settles rather than anything I did.
I also look fuller than I did a fortnight ago, and enough people have said so unprompted that it is not just me looking in a mirror. Two asked whether I train calisthenics, which is a fair read of 22 pull ups and three sets of 20 dips at 78.7kg. One put it down to genetics.
None of it is two weeks of new muscle. Tissue does not build that fast in anybody. In order of how much each is contributing, it is creatine pulling water into the muscle cell, glycogen resaturating now that the muscle gets emptied and refilled six days a week instead of one, and roughly double the weekly volume per pattern keeping everything in a state it now rarely leaves. Underneath those three, the size lost over the layoff is coming back, and regaining size you already had is far quicker than building it the first time. Bodyweight is still 1kg under March, which is the honest measure of where that stands.
The genetics comment is the interesting one. What is visible is the current state. What is not visible is seventeen years, a diet that has not changed in years, and a training week that runs six days out of seven. That reads as luck because the timeline is not in the room.
Monday, Push A
| Exercise | Performance |
|---|---|
| Bench press top set | 100kg × 5 |
| Bench press backoff | 85kg × 8, 8, 6, 6 |
| Standing shoulder press | 50kg × 8, 8, 6 |
| Incline dumbbell press | 32kg × 8, 8, 6 |
| Chest fly | 66kg × 15 for 3 sets |
| Cable lateral raise | 20kg × 15 for 3 sets |
| Triceps pushdown | 80kg × 15, 15, 12 |
| Dips | Bodyweight × 20 for 3 sets |
100kg for 5. Same bar, two more reps, seven days apart. Nothing was added in between and the load did not move, so this is the neural side coming back on schedule rather than anything the programme built. March was 100kg for 8 to 10. A week of training closed a third of that gap, which is what a strength loss rather than a tissue loss looks like when it reverses.
The backoffs went 8, 8, 6, 6 against 8, 8, 6, 5 last week. One rep, and it arrived on the set that needed it. Every set is now at 6 or above, which was the exact condition written down after the first Push A. So 85kg stays where it is and the next target is 7 across the back half, then 8, then 90kg.
Everything else pressing moved the right way. Standing press went 8, 8, 6 against 8, 6, 5. Incline went 8, 8, 6 against 8, 8, 5. Two full sets before the drop instead of one, on both. The collapse across sets is still there, it is just arriving later, and that is the fatigue signature receding rather than disappearing.
Dips came back in at three sets of 20. Twenty reps at bodyweight is conditioning and not a pressing stimulus, and it has the same problem the pull ups have, which is no way to load it. Until there is a dip belt they sit at the end of the session as volume and nothing more.
Chest fly reached 15 and lateral raises held 15 across three sets. Both at the ceiling of their range, so both go up next Push A.
Tuesday, Pull A
| Exercise | Performance |
|---|---|
| Bent over row top set | 105kg × 6 |
| Bent over row backoff | 85kg × 8 for 4 sets |
| Pull ups | Bodyweight × 8 for 4 sets |
| Lat pulldown | 59kg × 15, 12, 12 |
| Chest supported row | 25kg × 15, 12, 12 |
| Shrug | 40kg × 15, 12, 12 |
| Face pull | 50kg × 15 for 3 sets |
| Hammer curl | 14kg × 12 for 3 sets |
105kg for 6 on the top set and 85kg for 8 across all four backoffs. Both went exactly as written, which makes this the first session in the block where the prescribed loads were neither too light nor too heavy. Row is 5kg ahead of where it opened on both, the backoffs are still holding every set, so it goes to 90kg next Tuesday.
I also timed the rest for the first time. Two minutes on the backoffs, ninety seconds on pull ups, sixty on the isolation work, with a clock running instead of a feel for it. Seventeen years of training and I have never done that once. Left alone I start a set when I am fully recovered, which is sometimes correct and more often just long.
It changed what the numbers say. Lat pulldown went 15, 12, 12. Chest supported row went 15, 12, 12. Shrugs went 15, 12, 12. Three exercises, the same shape on all of them, a first set clear of the range and two more settling under it. Under the old habit those would have read 15, 15, 15, because the rest would have stretched until they did. So three flat sets at one weight was never evidence the weight was right. It was evidence I waited long enough.
The correction is upward regardless. A first set of 15 against a range of 10 to 12 means 59kg is light, so the pulldown goes to 66kg. Face pulls held 15 across three sets and go up. Hammer curls stay at 14kg until all three sets reach 15.
Weekly Log
Primary Lifts
| Lift | W1 | W2 | W3 | W4 | W5 | W6 |
|---|---|---|---|---|---|---|
| Bench press | 100kg × 3 | 100kg × 5 | ||||
| Bent over row | 100kg × 6 | 105kg × 6 | ||||
| Shoulder press | 60kg × 6 | |||||
| Squat | 130kg × 4 | |||||
| Deadlift | 160kg × 3 | |||||
| Pull ups | 22 |
Bodyweight
| Week | Bodyweight (kg) | Notes |
|---|---|---|
| Pre block | 77.3 | Post layoff, before creatine |
| W1 | 79.0 | After 7 day creatine load |
| W2 | 78.7 | Load finished, 0.3kg off Week 1 |
| W3 | ||
| W4 | ||
| W5 | ||
| W6 |
Weekly Notes
Week 1
Sessions completed: 6 of 6
Volume raised from six exercises per session to eight before starting. Set order corrected to top set first, which is the single biggest change from how I have trained for seventeen years, and both the squat and the pull ups proved why inside the same week.
Pressing is clearly further behind than everything else. Bench backoffs fell apart by the fourth set on both push days while row and squat backoffs held every set comfortably at the prescribed weight.
| Pattern | Backoff verdict |
|---|---|
| Push | 85kg bench fell to 5 by set four, hold the weight |
| Pull | 80kg row held 8 across four sets, go to 85kg |
| Legs | 110kg squat held 8 across four sets, go to 120kg |
Pressing is the only pattern where the prescribed load turned out to be correct. That fits what the layoff did, since bench is where the drop showed up first.
The B sessions said the same thing louder. Rows at 70kg, squats at 100kg, leg curls, calf raises, face pulls and chest supported rows all finished at the top of their rep range without a drop. Six sessions in, almost every number outside pressing was written too low, which is the correct error to make in a first week and an expensive one to keep making in a second.
Deadlift enters the log at 160kg. That is a starting number rather than a measurement, and it is the only primary lift in the block with no honest baseline yet.
Started 3 August 2026. Updated weekly.