Most fitness apps claim to be "science-based." Few will tell you which science. This page does. Every rule our program generator enforces is documented here, traced to peer-reviewed research, and linked to the original publication.
If a finding evolves, the framework evolves with it. This document is versioned — the current revision is v2026.1, and prior versions are archived.
Meso organizes training into mesocycles: 3–5 week phases stacked in sequence — accumulation, then intensification, then peak, then deload — rather than training all qualities at once. A complete mesocycle runs roughly 9–15 weeks end-to-end, with phase lengths set at generation time based on your goal and training status.
This is the block periodization model, developed in Soviet sport science and systematized by Issurin as an alternative to traditional "mixed" periodization for higher-level athletes. The four-phase structure used here follows the strength-training periodization synthesis developed in parallel to Issurin's framework (Plisk & Stone 2003); Issurin's own taxonomy is three-block (accumulation, transmutation, realization), and the terminal deload week reflects the hypertrophy-training convention covered in Pillar 06. The core argument: concurrent development of many abilities dilutes the stimulus for each; consecutive development of a few abilities at a time produces cleaner, larger adaptations. Bartolomei et al. (2014) tested this structure directly against a traditional within-mesocycle program in experienced lifters and found a "possibly to likely" advantage for the block arm on upper-body strength and power.
The strongest evidence base for block periodization is in competitive athletes preparing for peak-performance events; transfer to recreational hypertrophy-focused training is reasonable but less directly studied. BP is also not unambiguously superior to all alternatives — daily-undulating and traditional within-mesocycle variation produce comparable results in some populations and timeframes. Meso uses block periodization for its alignment with goal-specific phase emphasis and its compatibility with adaptive program generation.
3–5 weeks per phase, ~9–15 weeks total across the four phases. Phase proportions are set at generation time based on the user's goal (hypertrophy-leaning vs. strength-leaning) and their current training status.
Open the Phase explainer to see exactly which block you're in, what comes next, and why this week is shaped the way it is. The macro cycle isn't hidden — it's two taps from the home screen.
Fixed percentages of 1RM assume your strength, sleep, and recovery are constant. They aren't. Meso uses RPE (Rating of Perceived Exertion) anchored to RIR (Repetitions in Reserve) to let the target stay the same while the weight floats with your daily capacity.
An RPE 8 set means "I could have done 2 more reps" — regardless of whether today that's 142.5 kg or 135. Hackett and colleagues (2012) introduced repetition-to-failure estimation in resistance training and showed lifters can predict their remaining reps with sub-one-rep accuracy across bench press and squat (r = 0.93–0.95 between estimated and actual repetitions to failure). Zourdos and colleagues (2016) formalised the specific 1–10 RIR-anchored scale Meso uses and validated it against movement velocity in trained squatters.
Two caveats worth surfacing. The scale's direct validation evidence comes from free-weight compounds (bench press and squat); application to isolation and machine-based exercises follows the same construct but rests on a thinner empirical base. And RIR estimation is most reliable at RPE 7 and above — lower values (RPE 5–6, used for deload) function as a ceiling rather than a precise target.
RPE (6–10). When you log reps + RPE, Meso adjusts the next week's load by a scheme-specific rule — rather than adding fixed % increments that may or may not match your current state.
Every set on the Today screen carries its target weight and an RPE cap. Hit the cap a rep early and the next session adjusts down; finish with reps in reserve and the next target steps up. The autoregulation rule is visible, not buried.
More volume is not infinitely better. Each muscle has a floor below which training doesn't grow it (MEV — Minimum Effective Volume), a range where growth is efficient (MAV — Maximum Adaptive Volume), and a ceiling beyond which recovery fails (MRV — Maximum Recoverable Volume).
These landmarks — measured in working sets per muscle per week — synthesize findings from Schoenfeld and colleagues on dose-response to resistance training, and were systematized for practitioners by the Renaissance Periodization group (Israetel et al. 2021). The 12–20 sets/week adaptive range is best-supported in young trained men aged 18–35; thresholds for other populations are extrapolated from the volume-landmark framework.
volume audit on every generated program. Accumulation sits near MAV; deload drops below MEV. Under-MEV violations are surfaced as audit findings and corrected on retry; above-MRV is a calibration nudge rather than a hard block — Schoenfeld 2017 explicitly states the upper limit of the dose-response curve is unknown, and Baz-Valle 2022 shows triceps benefit past 20 sets/week.
A growing body of research suggests that training a muscle in a lengthened position — where it's mechanically under stretch — directionally favours greater hypertrophy than training it in a shortened position, at equated volume and effort.
The meta-analysis by Wolf et al. (2023) directionally favoured full range of motion over partial ROM, though the effect sizes are trivial-to-small in the meta-analytic average (SMD 0.04–0.12 across outcome categories). Well-designed individual studies that specifically manipulate muscle length — Maeo et al. (2022) on the triceps, with a medium effect (d = 0.54) at equated effort — point to long-muscle-length training as the driver. Meso uses this at the exercise-selection stage: movements are tagged as stretch-biased or not, and the generator prefers stretch-biased options for each muscle group.
is_stretch_biased boolean. The microcycle audit surfaces missing stretch-bias coverage per muscle as a low-severity calibration note rather than a hard block — the literature directionally supports the preference but doesn't underwrite a structural rule. The injury profile can still rule out specific stretch-biased options.
Schoenfeld and colleagues' 2019 meta-analysis — the largest synthesis available on this question, with 25 studies and ~800 subjects — found that when total weekly volume is held constant, weekly frequency per muscle does not meaningfully affect hypertrophy. Training a muscle once, twice, or six times per week produces comparable growth, provided the same number of working sets is performed. The earlier 2016 meta-analysis (10 studies, ~200 subjects) had favoured 2×/week; the expanded evidence base revised that finding. When volume is allowed to scale with frequency, higher-frequency groups show a modest advantage — which the authors interpret as frequency functioning as a tool to accumulate volume, not as an independent stimulus.
Meso defaults to programs that train every muscle at least twice per week. This is an operational choice, not a science-derived minimum — the 2019 evidence shows 1× can produce real hypertrophy at matched volume. The default distributes recovery load more evenly across the week and supports more granular progression feedback per exercise. The 3-day Push/Pull/Legs split is the one inherent exception, training each muscle once per week, which the engine accepts as a known tradeoff.
days_per_week intake — one default split per day count, not a menu. auditFrequency flags muscles trained <2× per week as a low-severity calibration note rather than rejecting the program. The 3-day PPL split is treated as a known carve-out: each muscle is trained once per week, which the 2019 evidence supports as a viable configuration at equated volume.
Chronic accumulation of training stress without sufficient recovery produces what the literature calls non-functional overreaching — performance decrements that take weeks to fully resolve once accumulated, accompanied by mood disturbance and prolonged impaired recovery. (True overtraining syndrome — performance loss persisting for months — is rare in resistance-trained populations; Bell et al. 2020 found no documented cases in strength-sport literature.)
A deload — a planned reduction in training stress — allows accumulated fatigue to dissipate while preserving adaptive gains, producing the supercompensation response that underlies long-term progress. Meso schedules a deload at the end of every mesocycle. The cadence is structural — you don't have to ask for it — but the engine can pull a deload in earlier when your readiness signals say it's needed, consistent with Kreher's prevention guidance to adjust training against mood and performance.
4 ≤ weeks ≤ 6 of loading + 1 week deload. Deload week: volume reduced 40–50%, intensity reduced ~20%, RPE capped at 6. The supercompensation mechanism — volume and load reduction on the same exercises — is locked: edits during a deload that increase load, volume, or RPE are hard-blocked. Schedule shifts (travel weeks, length adjustments, volume-preserving exercise swaps) are allowed with a reason.
The Plan's 30-day calendar marks deload weeks ahead of time so you can see them coming. When the engine pulls one in early based on your readiness signals, you'll see why — not just a sudden lighter session.
Adding weight every week is one valid progression scheme — but only one of several, and not always the right one for a given exercise or phase. Meso assigns each exercise an explicit progression scheme drawn from the resistance-training literature, chosen to match movement type, phase, and lifter experience:
Linear — a fixed micro-load added per week (e.g., +2.5 kg/week on squats). Default for compound lifts in less-experienced lifters. Suchomel et al. (2021) describes linear loading as appropriate for novices, with the caveat that exclusive long-term use leads to stagnation.
Double progression — keep weight fixed, add reps within a target range, then jump weight and restart the rep range. Default for hypertrophy accessories where small plate jumps aren't available. Plotkin et al. (2022) shows rep progression is a viable hypertrophy stimulus comparable to load progression.
Wave loading — intensity undulates across weeks (e.g., 3 × 6, 4 × 4, 5 × 3, back down), letting volume and intensity both progress across multiple weeks. Default during intensification phases. Variation in intensity across multi-week blocks is consolidated by Suchomel et al. (2021) as a fatigue-management strategy.
RPE-based — load is not prescribed in advance; it's derived from the previous session's RPE. Default during deload phases and for advanced lifters whose readiness varies day-to-day. Helms et al. (2016) is the foundational practitioner reference; Suchomel et al. (2021) consolidates the broader autoregulation literature.
Recent direct experimental evidence (Plotkin et al. 2022) shows both adding load and adding repetitions produce comparable hypertrophy over an 8-week block in trained lifters — meaning the choice of progression mechanism matters less than the presence of progression itself.
progression_scheme ∈ {linear, double, wave, rpe}. Assignment is deterministic, based on movement type (compound vs. isolation), current phase, and the user's training status. Schemes do not change mid-cycle except when audit triggers a retry.
The History heatmap shows training density at a glance, with PR markers on each lift and session-to-session weight increments visible in the timeline. The progression rule isn't just a promise — it's a chart.
Within a session, the exercise you perform first is the one that gets the freshest neural drive, the highest loads, and the cleanest technique. Strength gains follow position cleanly: exercises performed first in a session produce greater strength improvements in those exercises over time (Nunes et al. 2021, meta-analytic specificity effect g = 0.45, p = 0.014). Hypertrophy is less position-sensitive — the same meta-analysis found no meaningful effect of ordering on muscle growth across seven studies (g = 0.03, p = 0.862), with the caveat that measurement methods in most included studies couldn't separate primary-mover from synergist hypertrophy cleanly.
Compound multi-joint movements (squat, deadlift, bench press, overhead press, row, pull-up) are the typical opener for three converging reasons: they recruit the largest absolute loads, they place the highest technical demand on still-fresh systems, and they generate fatigue that's most disruptive to subsequent compound performance. Isolation work typically follows. Where a lifter's priority is a specific muscle without a dedicated compound — say, side-delt or biceps development — the priority exercise can lead instead.
The optimal training dose shifts upward with experience. Untrained lifters reach their maximum strength gains around 60% 1RM and roughly 4 sets per muscle; trained lifters need closer to 80% 1RM at the same volume; competitive athletes peak around 85% 1RM at roughly double that volume (Rhea et al. 2003; Peterson et al. 2004). The same program that pushes a novice past their optimal point leaves an advanced lifter short of theirs.
The American College of Sports Medicine's progression-model framework (Kraemer et al. 2002) structures these adjustments around three training-status tiers — novice, intermediate, and advanced — each with distinct loading, volume, and frequency targets. Within-population variation compounds the picture: in a 12-week study, lifters running the same program ranged from no measurable strength or muscle gain to 250% strength increases and ~60% hypertrophy (Helms et al. 2020).
Meso asks for your training status at intake — novice, intermediate, or advanced — and uses it as a global multiplier on starting volume, progression increments, MRV, and phase length. The strongest dose-response evidence stratified by training status comes from strength outcomes (1RM gains) in young trained men and competitive athletes; Meso applies the same training-status modulation across both strength and hypertrophy targets — directionally supported by the cited literature, with magnitudes calibrated from practice.
training_status sets the phase-length range, starting volume multiplier, and default progression aggressiveness. A novice's linear progression might add 2.5 kg/week; an advanced lifter's might be 1 kg/month on the same lift.
Sports science doesn't stand still. Findings are refined, meta-analyses are re-run, and our framework updates accordingly. Every change to this document is archived with a version number and a dated changelog.
If you spot a rule we've missed, a source we should cite, or a study that supersedes one of ours — research@meso.app. We read every email.
Disclosure on cited authors. Several papers cited across these pillars share overlapping authorship. Brad Schoenfeld appears as lead or co-author across Pillars 04, 05, 07, and 08; Eric Helms and Michael Zourdos collaborate across Pillars 02, 07, and 09; the Renaissance Periodization framework underwrites Pillar 03's volume-landmark terminology and funding on papers cited in Pillars 04 and 07. Methodology and findings have been audited on their merits — and where independent research groups have validated the same construct (Hackett 2012 alongside Zourdos 2016 on RIR estimation), we cite both.