SYNTH-75 · Guide
The first 12 weeks
Build the habit before the session: frequency first, then duration, then one short interval session; a beginner improves on almost anything, so the plan is limited by tissues and adherence, not by the stimulus, and a fit cyclist who starts running is a beginner below the hips
The question it answers
"I am starting from little or nothing — how many days, how long, how hard, do I need intervals, when do I test, when do I lift, what should I expect?" and "I am a fit cyclist starting to run — how much can I run?"
Default (what to prescribe first)
Not from a study; the base's own practice (order and numbers): frequency → duration → intensity.
1
Weeks 1-4: 3 sessions a week on non-consecutive days, 20-40 min, all conversational. Cycling: continuous easy riding. Running: run-walk (below). The goal is a week the athlete repeats, not a hard session.
Basis for frequency first (inference; the order itself is untested): in a meta-regression of 5,973 participants, more sessions per week (6 > 4 > 2) and lower initial fitness both went with larger mitochondrial gains, and endurance, HIT and sprint training gave similar gains once frequency, weeks and initial fitness were adjusted (REF-465, A; populations from diseased and untrained to trained).
2
Weeks 5-8: add a 4th day or lengthen sessions — one change at a time; grow one session into a longer easy one. In previously inactive adults the VO2max gain from continuous moderate training kept rising with session length (80% of its observed maximum at 52.4 min a session, not flattened at the longest dose studied) (REF-749, A, 69 RCTs, n=2387, Tier 0-1).
3
Weeks 5-12 (cyclists from week 3-4): at most one short interval session a week, optional. In the same meta-analysis interval training raised VO2max more than continuous (d = 0.38, 1.20 mL/kg/min), reached 80% of its gain at 11.4 min a session including recoveries (95% CI 9.5-40.2) and added little beyond ~23 min (REF-749).
Practice: e.g. 5-6 × 1 min hard / 1 min easy after a warm-up (SYNTH-66; that framework notes beginners are untested); cyclists may start it from week 3-4, runners only once they run 20-30 min continuously without symptoms.
4
Everything else stays easy enough to speak in full sentences. In 16 healthy volunteers on a treadmill and a cycle ergometer, the stage where speech first became difficult did not differ significantly from the ventilatory threshold, and once speech was not comfortable intensity was consistently above it (REF-929, B, incremental test, no limits of agreement).
Decision
| Situation | Do this | Source and how sure |
|---|---|---|
| True beginner, cycling | 3 → 4 days; easy continuous; one short interval set from week 3-4 at the earliest | REF-749 A, REF-465 A (direction); days and weeks = practice |
| True beginner, running | 3 days, run-walk, time not distance; lengthen the run parts before adding a day; no fast running until continuous running is symptom-free | Practice. Why tissue is assumed to set the pace (inference; the review counts injuries and does not explain them): novices were injured 17.8 times per 1000 h of running against 7.7 in recreational runners (REF-928, A-, 13 studies, differing injury definitions; counts, not causes). Run-walk itself is untested in beginners here: the only trial is 42 non-elite marathoners, where run/walk cost no significant time (4:14:25 vs 4:07:40, p = 0.377) and left less muscle pain and fatigue (REF-894, B+; no run:walk ratio in the abstract) — a transfer from race day to training |
| "Do I need intervals?" (Tier 0-1) | No, not to improve; yes if time is short. One short set is a complete session | REF-749 A: both modes work (within-group d 0.83 interval, 0.48 continuous). In 45 inactive women, 12 weeks of self-paced cycling 3×/week raised VO2peak 16 ± 8% (intervals) and 21 ± 12% (continuous), enjoyment similar (REF-242, B+). In 29 sedentary young adults, three 10-min sessions with one or two 10-20 s sprints raised VO2max 15% (men) / 12% (women) in 6 weeks (REF-051, B) |
| How fast to add running volume | Keep week-to-week increases under ~20%; never jump by a third or more; symptoms, not a percentage, are the gate (practice) | REF-623 B+: 261 recreational runners, 20-60% jumps → 22.6 points more injuries at 21 days (95% CI 0.9-44.3), no difference at 56 and 98 days. REF-565 B+, 4 studies: 10% vs 24% weekly increase, same risk (HR 0.8, 0.6-1.3) — no "10% rule". REF-925 A, the one RCT: 532 novices, a graded 13-week programme built on the 10% rule vs a standard 8-week one, injuries 20.8% vs 20.3% (P = .90). REF-619 B: 874 novices, >30% over 2 weeks vs <10%, HR 1.59 (0.96-2.66, p = .07) for distance-related injuries only; overall rate not different |
| Zones before any test | Talk test and RPE: full sentences, roughly RPE ≤13 on the 6-20 scale = easy. No %HRmax from an age formula | Talk test: REF-929 B (n=16, healthy volunteers, running and cycling); REF-930 C+ (narrative review; may not be practical for high-intensity interval training). The RPE figure is practice (SYNTH-51, SYNTH-40). Every age formula carried RMSE 9.1-10.5 bpm in 5069 athletes (abstract count; 5311 in the Results, corrected 2026-10-12) (REF-3494, B+; 83.5-89.7% male, Tier 2-3 — transfer to beginners is inference). Fixed %HRmax has little merit for placing someone in a domain (REF-173, B+, narrative) |
| First test | Practice: none in weeks 1-4; a first field test around weeks 6-8, once the athlete can pace a sustained effort; label its protocol (SYNTH-53, SYNTH-38) | No entry tests timing. A single before/after test cannot separate a real response from day-to-day variation (REF-430, B+, methodological) |
| Strength | From week 1 if the athlete wants it: 2-4 weeks at moderate loads to learn the lifts, then SYNTH-44; not heavy legs the day before the longest session | Row = practice (SYNTH-44). Across 25 RCTs in general sport, strength programmes cut injuries (RR 0.315, 0.207-0.480; overuse RR 0.527) and stretching did not (RR 0.963) (REF-861, A-; not endurance-specific). The one trial in runners is null: 720 first-time marathon runners, a self-directed 10-minute programme 3 times weekly vs observation, overuse injury resulting in noncompletion 7.1% vs 7.3% (REF-924, A-; light, unsupervised). So offer strength for its own sake, not as injury insurance. Untrained muscle is damaged more by the same bout: 82% vs 45% of fibres disrupted in the eccentric arm (REF-702, B; arm-curl model, two separate studies) |
| Fit cyclist adding running (Tier 2-3 on the bike) | Cap by tissue, not by breathing: start as a beginner runner (3 short run-walk or easy runs a week), same <20% rule, no fast running, hills or long runs in the first 6-8 weeks; keep the engine work on the bike | Practice — no entry studies this athlete. See next section for the reasons |
Population (b): the cyclist who starts to run
The engine is not the limiter, so effort is a false guide. Cyclists reach a similar VO2max cycling and running, and threshold adaptations are specific to the trained mode (REF-537, B+, narrative review). Inference: easy running will feel trivially easy at a pace the legs have never been loaded at.
Bone: adult road cyclists in regular training have low bone mineral density at key sites such as the lumbar spine, and road cycling does not appear to build bone (REF-896, A-, 31 studies, no pooling).A bone stress injury is a workload error — number × magnitude of loading cycles beyond the bone's tolerance — prevented by sensible progression, particularly of intensity (REF-622, C+, narrative). "Progress running slowly from a non-impact background" is the base's practice in REF-522 (C+), not a tested protocol.
Tendon: tendon stiffness rose in loading interventions of ≥8 weeks in healthy adults aged 18-50 (SMD 0.70), and depended on load intensity (REF-517, A; strength interventions, not running). Inference: the first 8-12 weeks of running are the tendon's adaptation window whatever the heart rate says.
Muscle: eccentric damage tracks mechanical strain, not metabolic effort, and a second bout a week later produces much less damage (REF-704, B+, narrative). Inference: expect soreness out of proportion to effort after the first runs, then a quick drop; do not read the drop as tissues being ready.
Keep the fitness on the bike. In four elite triathletes cycling training carried over to running performance (r = 0.56) (REF-785, B, n=4, modelled). Inference: replace bike hours with running slowly; do not add running on top of a full bike week and a hard block at once.
Heart rate differs between modes at maximal and submaximal intensities (REF-537): set run HR zones separately (inference in that entry), never copy bike zones.
Practice, not evidence: first weeks 3 × 15-25 min easy or run-walk on flat, even ground, never on consecutive days; first continuous 30 min before any quality; first run intervals not before weeks 8-12.
How to progress / periodise
One variable at a time (a day, or minutes, or the interval set) — practice; ramp and deload signals are SYNTH-29.
Run-walk progression — practice, no entry gives a ratio: e.g. 1 min run / 2 min walk × 8-10, lengthening the run part every 1-2 weeks while the previous week left no pain that lasted into the next day, toward 20-30 min continuous by weeks 8-12. Repeat a week rather than force the step.
Missed a week: restart one step back (practice; with an established base, SYNTH-36).
What to expect. Large and variable: in 481 previously sedentary adults, 20 weeks of standardised training raised VO2max by about 400 ml/min on average, with some gaining little or nothing and others over 1.0 l/min (REF-429, B+, single-arm).A low response in one block does not predict the next: in 42 untrained middle-aged adults repeating the same 8-week programme, individual VO2max response reproduced poorly (ICC 0.19) (REF-498, B+). So never label a beginner a non-responder from the first 12 weeks.A large gain at Tier 0-1 may be a novice effect that does not transfer upward (REF-529, an inference in that entry).
When NOT to use it / red flags
Every session medium-hard. In 30 recreational runners over 10 weeks, the easy-dominated polarized programme improved 10 km by 5.0% vs 3.6% for the between-thresholds one — not significant in the full sample, clearer only in the 12 who kept easy days easy (REF-086, B+; Tier 1-2, 10-km times ~39 min, not true beginners).In 20 trained amateur runners, easy runs at the first threshold gained nothing over 10-15% below it (REF-705, B+). Inference for beginners: harder easy days buy nothing shown and cost recovery.
Longer interval sets for a beginner: the curve is flat past ~23 min a session (REF-749); add an easy session instead.
Localised bony tenderness, pain that worsens through a run or is present at rest or at night: stop running and refer — the base's practice in REF-622; do not coach through it.
Previous injury is a strong-evidence risk factor in long-distance runners (REF-564, A-): slowest ramp at that site. In 930 novice runners a previous running-related injury was not a predictor (P = .47); age between 45 and 65 years, BMI >30 kg/m² and previous injuries not related to running went with more injuries — weak signals, mostly not significant (REF-926, B+, observational); reasons for care, not rules.
Medical clearance before starting is outside this base; with symptoms on exertion, a doctor comes first.
What the evidence does and does not show
Shows (A, Tier 0-1): both continuous and interval training raise VO2max in inactive adults; the interval dose saturates at short sessions, the continuous dose does not (REF-749). The outcome is VO2max, not performance, and "interval" pools near-maximal and all-out protocols.
Does not show: that "frequency before duration before intensity" beats another order; that run-walk prevents injury in beginners; any dose for a cyclist starting to run; when a beginner should first test. All four are practice here.
Evidence gaps
No trial of larger weekly jumps, or of any progression rule other than the 10% rule (REF-925), in novice runners.
Talk test: no limits of agreement, no test in long sessions (REF-929 is one incremental test).
Supervised or heavy strength training as injury prevention in new runners (REF-924 was a light home programme).
No study of cyclists or other aerobically fit non-runners starting to run (injury, bone or tendon response).
No beginner run-walk training trial, and no run:walk ratio anywhere in the base.
No performance time course for Tier 0-1 over 12 weeks; no adherence data by programme design.
No study of how often or how early a beginner should be tested.
How this entry was checked and changed
Why this level
(the VO2max dose rests on an A dose-response meta-analysis in inactive adults REF-749 and an A meta-regression REF-465; running progression on B/B+ observational cohorts REF-619 REF-623 a B+ review of four studies REF-565 and a null A RCT of the 10% rule REF-925; the talk test on one small B validation REF-929; the 12-week structure, run-walk progression and every rule for the cyclist-turned-runner are house practice with no trial behind them)
Content correction 2026-09-27
Revision 2026-10-01: folded REF-925 (Buist 2007, 532 novices: graded 10%-rule programme vs standard, 20.8% vs 20.3% injured) into the progression row, the lede and the level line. 'No support for a 10% rule' now rests on a randomised trial as well as the review. The practice lines 'under ~20% a week' and 'never jump by a third' are unchanged and stay with REF-623 and REF-619 — REF-925 does not test larger jumps.
Revision 2026-10-01: folded REF-929 and REF-930 (talk test). 'Easy enough to speak in full sentences' and the talk-test half of the zones row lose the practice label and cite REF-929 (n=16, healthy volunteers, incremental test) and REF-930 (narrative review); the RPE figure stays practice.
Revision 2026-10-01: folded REF-928 (novices 17.8 vs recreational 7.7 injuries per 1000 h) as the reason tissue sets the pace, and REF-926 (930 novices: previous running injury not a predictor; age, BMI, non-running injuries weak signals) next to the previous-injury red flag, which rested on mixed long-distance runners.
Revision 2026-10-01: REF-924 (720 first-time marathoners, self-directed strength, null) added to the strength row. The row cited a general-sport meta-analysis as if strength from week 1 prevented injury; the prescription is unchanged (strength from week 1 if the athlete wants it) but it is no longer offered as injury insurance for a new runner. Three Evidence-gap lines narrowed, one added. Level unchanged at B-: the 12-week structure is still house practice.
Verification 2026-10-01: all 32 cited entries opened in the KB.md snapshot; levels, designs, n and numbers next to each refId match (5,973; 6>4>2; 52.4 min; 69 RCTs/2387; d 0.38/1.20; 11.4 min, CI 9.5-40.2; ~23 min; d 0.83/0.48; n=16; 17.8 vs 7.7; 42, 4:14:25 vs 4:07:40, p 0.377; 45, 16±8/21±12 %; 29, 15/12 %; 261, 22.6, 0.9-44.3; HR 0.8, 0.6-1.3; 532, 20.8 vs 20.3 %, P .90; 874, HR 1.59; RMSE 9.1-10.5, 5069; 25 RCTs, 0.315/0.527/0.963; 720, 7.1 vs 7.3 %; 82 vs 45 %; 31 studies; SMD 0.70; r 0.56; 481, ~400 ml/min; ICC 0.19; 5.0 vs 3.6 %, 12; n=20; 930, P .47).
Verification 2026-10-01: Level B- is fair: 2 of 4 Default steps rest directly on REF-749 (A, Tier 0-1) and REF-929 (B), the other two mix a cited basis with labelled practice. No LEVEL-OVERALL finding.
Verification 2026-10-01: no unlabelled prescription found beyond the one in the 'Too much too soon' line; the 12-week structure, run-walk progression, test timing, strength row and every cyclist-to-runner rule carry a practice or inference label. The title's 'limited by tissues and adherence, not by the stimulus' is itself inference and unlabelled in the title; the lead labels the untested parts.
Verification 2026-10-01: no SOURCE flags: REF-929 and REF-498 have no DOI but have a PMID.