Can one hard week sharpen me before a race?
In 19 competitive male cyclists, 7 days of ~2-h sprint sessions, then a light week, raised 20-km TT power 4.6-6.8% by day 14; controls fell 3.3%. Not randomised.
Clark and colleagues tested a practice common before important races: a short period of overload with high-intensity training.
Twenty-eight competitive male cyclists, in season, were matched on peak power and VO2peak and assigned to a control group (n = 9) that kept its usual training (at least 10 h a week) or to one of two groups doing seven consecutive days of about 120-min sessions of maximal sprints: short (25 sets of 5, 10 and 20 s; n = 9) or long (10 sets of 15, 30 and 45 s; n = 10), at a 1:5 work-to-rest ratio and about 15 min of total sprinting a session.
Both groups then rode only light recovery training, with no hard sessions or races, for 7 days, and were tested 7 and 14 days after the block on a 20-km hilly time trial and an incremental test. At 14 days, time-trial power relative to the control group was 8.2% ± 3.8% (short) and 10.4% ± 4.3% (long) higher (± 90% confidence limits), and peak power 5.5% and 9.5% higher.
Within groups, time-trial power rose 4.6% and 6.8% while the control's fell 3.3%. VO2peak, power at 4 mmol/L lactate and gross efficiency rose non-significantly within groups; relative to control, the power at 4 mmol/L was significantly higher (+7.5% and +6.8%). The short-sprint group showed no time-trial gain at 7 days and needed 14.
Our reading: a hard, race-like week followed by an easy one sharpened 20-km power within two weeks in competitive amateurs — a package of overload plus recovery set against continued normal training, not a test of block against traditional periodization at equal load, although the meta-analysis REF-946 counts it as one.
Protocol
Participants: 30 volunteered; 2 withdrew with illness unrelated to the study; n = 28 men, 33 ± 10 y, 74 ± 7 kg, 178 ± 5 cm, VO2peak 4.7 ± 0.5 L/min; grade A or B (Oceania amateur grading), at least two years of competition; tested in the competitive season after base and pre-competition training; at least 10 h (about 300 km) of mixed-intensity training a week in the 2 weeks before. Prior training was not controlled.
Assignment: matched as closely as possible on peak power and VO2peak from the first test, then assigned to control (n = 9), short HIT (n = 9) or long HIT (n = 10). Not randomised.
Intervention: seven consecutive days; each session about 120 min including a self-selected 15-min warm-up and cool-down. Short: 25 sets of 5, 10 and 20 s (14.6 min of work, 73 min of recovery). Long: 10 sets of 15, 30 and 45 s (15 min of work, 75 min of recovery).Work-to-rest 1:5, efforts self-paced at the highest possible intensity, recovery at about 30-40% of peak power, timed by audio cues. Sessions 1, 4 and 7 supervised in the laboratory; the rest on the riders' own bikes, on the road or a stationary trainer. Compliance 100%.
After the block: light recovery training (under 120 min) and no high-intensity exercise or racing for the 7 days immediately after; what was done on days 8-14 is not described.
Control: own training, at least 10 h a week, "to ensure that total training volume was similar"; tested twice, 3 weeks apart. Its training was not reported.
Tests (Velotron ergometer, about 20°C): incremental test from 100 W, +40 W every 4 min, for peak power, VO2peak, power at 4 mmol/L lactate (OBLA), economy and gross efficiency at 220 W; 20-km simulated hilly time trial, coefficient of variation about 1% for time and 2% for power (the group's in-press data). The HIT groups were tested twice before (averaged) and 7 and 14 days after the block.
Statistics: repeated-measures ANOVA, pre to the 14-day test, Bonferroni-corrected pairwise comparisons; magnitude-based inference with 90% confidence limits; smallest worthwhile change 1% for power.
Funding and interests: a CAPES travel scholarship for one author; the funders had no role; no competing interests declared.
Key Results
Pre to 14 days after the block (Tables 1 and 2; within-group mean ± SD; between-group difference ± 90% confidence limits):
| Measure | Control (n = 9) | Short HIT (n = 9) | Long HIT (n = 10) | Short vs control | Long vs control |
|---|---|---|---|---|---|
| 20-km TT power | −3.3 ± 4.2% | +4.6 ± 4.4% | +6.8 ± 5.8% | +8.2 ± 3.8% (p < 0.01) | +10.4 ± 4.3% (p < 0.05) |
| 20-km TT time | +1.8 ± 2.2% | −2.9 ± 2.6% | −4.4 ± 3.7% | −4.6 ± 1.9% | −6.1 ± 2.2% |
| Peak power (incremental) | −1.7 ± 3.3% | +3.6 ± 3.0% | +7.6 ± 2.3% | +5.5 ± 2.7% | +9.5 ± 2.5% |
| VO2peak | −0.6 ± 6.3% | +2.3 ± 4.8% | +3.5 ± 6.2% | +2.9 ± 4.6% (ns) | +4.2 ± 5.1% (ns) |
| Power at 4 mmol/L (OBLA) | −3.6 ± 6.4% | +3.6 ± 3.5% | +2.9 ± 5.3% | +7.5 ± 4.5% (p < 0.05) | +6.8 ± 4.9% (p < 0.05) |
| Gross efficiency at 220 W | +1.5 ± 4.3% | +3.2 ± 2.4% | +5.1 ± 3.9% | +1.7 ± 2.9% | +3.6 ± 3.3% |
Short vs long: peak power favoured the long efforts (difference 3.7 ± 2.1%, p < 0.05); time-trial power differed by 2.1 ± 3.9% (not significant).
Time course (Fig. 3): the long group's time-trial gain was present at 7 days; the short group showed no time-trial improvement at 7 days and improved by 14 days.
Abstract and body disagree: the abstract and conclusion call the physiological changes non-significant, but relative to control the power at 4 mmol/L was significantly higher in both groups (Table 2; pairwise p = 0.024 and 0.034 in the text).
Significance marks that do not add up: the text lists the short-vs-control time-trial-time comparison as significant at p = 0.06, while Table 2 marks it p < 0.01; Table 2 marks long-vs-control time-trial power only "*" (p < 0.05) although the text gives p < 0.001 (under-marked, not contradictory); and the abstract's "significant (p<0.05) gains in time trial performance" for both groups does not hold for the short group's time-trial time at the text's p = 0.06.The abstract gives the short group's VO2peak change as 2.3% ± 4.7%, Table 1 as 2.3 ± 4.8.
Size, honestly: the "8-10% relative to control" includes the control's own 3.3% fall over 3 weeks of in-season training; the riders' gain over their own baseline was 4.6-6.8%.
Mechanisms
Measured: relative to control, power at 4 mmol/L rose about 7% in both groups; VO2peak, economy and gross efficiency changes were not significant; peak power rose more with the longer efforts.
Authors' interpretation: the long-sprint gains went with better economy (4.6%) and gross efficiency (5.1%), the short-sprint gains with power at 4 mmol/L (3.6%); improved muscle buffering is possible but was not measured. The short sprints were probably harder (mean power about 10% higher in unpublished case observations), leaving residual fatigue that delayed the benefit; they frame the block as functional overreaching followed by supercompensation.
Inference: the HIT groups were tested after a week or more of light training, while the control kept training at least 10 h a week; part of the difference may be a taper-like effect of reduced load after the block (REF-107), and part is the control's unexplained 3.3% fall. The design cannot separate the overload from the recovery that followed it.
Inference: the sessions were built from race power data (repeated maximal efforts of 5-45 s), so the block was specific to road racing as much as it was "more HIT".
Practical Application
Can one hard week sharpen me before a race? In 19 competitive male cyclists, 7 days of ~2-h sprint sessions, then a light week, raised 20-km TT power 4.6-6.8% by day 14; controls fell 3.3%. Not randomised.
What was done: seven days in a row of about 2 h each — sets of maximal 15, 30 and 45-s efforts (or 5, 10 and 20 s) with five times as long easy between, about 15 min of sprinting in total — then 7 days of light riding under 2 h with no hard sessions or races, and the test at day 14 (B, n = 28, Tier 2-3 men).
Allow two weeks; with the shortest, hardest efforts the time trial had not improved by day 7. The longer efforts had paid off by day 7; the shorter, more intense ones only by day 14. The authors advise longer efforts, or a longer recovery, before a race.
Read the size honestly. Over their own baseline the riders gained 4.6-6.8% in time-trial power; the 8-10% is relative to a control whose power fell 3.3%.
A short window may hold a short, race-specific overload (one non-randomised trial). Rested riders gained from a 7-day block within 14 days, against a control that declined.That qualifies the premise of the base's "no new build" line between two close races, "there is no time to absorb a new stimulus and taper it" (SYNTH-76), but the trial did not start after an A-race, and its sessions copied race efforts, closer to SYNTH-76's "sharpen" and to the specificity default (SYNTH-42) than to a new lever.
Limitations
Not randomised, small groups. Matched assignment, 9-10 per group; men only.
Matched for total volume at most, not for high-intensity work. The HIT groups added a 7-day overload and then a light week; the control continued normal training whose content was not reported. REF-946's study table describes the study as matched "by total and HIT volume", which only half fits this paper's methods: the control's training was set "to ensure that total training volume was similar" (its content was not reported), but nothing matched its HIT — the two HIT groups were matched with each other.
Overload and recovery are confounded. The post-tests followed 7-14 days of reduced load in the HIT groups only.
Partly unsupervised. Four of seven sessions were done on the riders' own bikes, on the road or a stationary trainer, without supervision; training on days 8-14 is not described.
Reporting. Significance marks in the text and Table 2 disagree for two comparisons; the abstract understates the lactate-threshold result; the time-trial reliability figures come from in-press data.
Laboratory time trial. A 20-km simulated course with the riders blinded to power; no race result.
Funding: a travel scholarship for one author; no competing interests declared.
How this entry was checked and changed
Why this level
not randomised (matched on peak power and VO2peak, then assigned), so outside the rubric's B+ band, which is for parallel RCTs of 20-39 (n = 28 in total, 9-10 per group); read from the full text (open access, CC BY 4.0)
Full text (checked 2026-10-26)
Twenty-eight male cyclists (mean ± SD, Age: 33±10 years, Mass 74±7 kg, VO2 peak 4.7±0.5 L·min−1); control group (n = 9), short sprint (n = 9) and long sprint (n = 10) HIT groups (full text)
Relative to the control group, time-trial power changed 8.2%±3.8% and 10.4%±4.3% for the short and long HIT regimes; peak power 5.5%±2.7% and 9.5%±2.5% (full text)
Within-group time-trial power: control −3.3±4.2, short 4.6±4.4, long 6.8±5.8 (% change, Table 1) (full text)
Power at OBLA relative to control: long 6.8±4.9, short 7.5±4.5, both marked significantly different at p<0.05 (Table 2) (full text)
Short group: 25 sets of 5, 10 and 20 seconds, 14.6 minutes of work and 73 minutes of recovery; long group: 10 sets of 15, 30 and 45 seconds, 15 minutes of work and 75 minutes of recovery (full text)
After the block: light recovery intensity training (<120 mins), no high intensity exercise or competitions in the 7 days immediately post HIT (full text)
Control: own training for a minimum of 10 hours per week; before the study all riders completed a minimum of ten hours (∼300-km) a week (full text)
Pairwise p for TT time, control vs short: p = 0.06 (text) (full text)
Competing interests: the authors have declared that no competing interests exist; CAPES travel scholarship for Dr. Costa (full text)