What should my last hard session before a race be?
In 8 trained 1500-m runners, 5 × 300 m at about 110% of race pace 3 days out, with continuous running cut 60%, went with a 5 s faster 1500 m; no direct test against race pace.
Spilsbury and colleagues compared two 7-day tapers, at least 3 weeks apart, in 8 highly trained male middle-distance runners, with a 1500-m time trial on an indoor 200-m track the day before and the day after each.
The race-pace taper (RP) followed prediction equations built from elite British runners' practice (REF-818): continuous running cut 30%, interval running cut 42%, and the final interval session — 5 × 300 m with 90 s recovery on day 5, three days before the trial — at 1500-m race pace.
The high-intensity taper (HI) was the same except that continuous running was cut 60% and the final session was prescribed at 115% of race speed; the runners actually ran it at 110% (43.8 s per 300 m against 41.0 s prescribed and 48.0 s in RP).
The 1500 m improved by 5.2 ± 3.7 s (90% confidence limits; 2.0%, p = 0.03) after HI, with 7 of 8 runners beyond the smallest meaningful change of 1%, and by 3.2 ± 3.8 s (1.2%, p = 0.15) after RP, with 5 of 8.
After both tapers the runners went out faster over the first 600 m and then held the same pace as before; one runner, the one with the highest training volume, got slower after both. The authors conclude that the low-volume, faster-final-session taper improves track performance more than the elite-practice taper, and advise trying both before minor races.
Our reading: both tapers improved most runners, and the faster, lower-volume taper did at least as well; with 8 runners, overlapping confidence intervals and no reported comparison between the tapers, there is no evidence of a difference between them, and the design cannot tell the deeper volume cut from the faster last session.
Protocol
Participants: n = 8 highly trained male middle-distance runners (800 and 1500 m); age 21.4 ± 4.2 y, height 182.8 ± 7.2 cm, body mass 67.4 ± 8.0 kg; 1500-m personal best 4:01.4 ± 0:09.2 min (range 3:47.6-4:11.7); at least two years of training and two uninterrupted months before the study.
Design: counterbalanced crossover, order by Latin square; two 7-day tapers separated by at least 3 weeks of their usual training; started in the 16th week of the winter training season; training otherwise set by their own coaches. Runners were not told the precise differences but could not be blinded to the change in load.
Baseline training (GPS, 4 weeks before; Table 2): continuous running 54 ± 14 km a week at 65 ± 4% of race speed, interval running 13 ± 4 km at 95 ± 3%, total 66 ± 16 km; 7 runs a week (5 continuous, 2 interval).
RP taper: prescribed from the elite-practice prediction equations: continuous 37 ± 8 km (−30%), interval 8 ± 2 km (−42%), total 45 ± 10 km (−32%); continuous runs 4 a week (−24%), interval sessions 2 (unchanged); final session at race pace.
HI taper: as RP except continuous running cut 60% (21 ± 5 km), total 29 ± 7 km (−56%), and the final session prescribed at 115% of race speed.
Final session (both): day 5 of the taper, three days before the post-taper trial; outdoor 400-m track, same time of day; warm-up of 10 min jogging, 10 min mobility and 5-6 strides; 5 × 300 m with 90 s recovery (1,500 m in total); prescribed 6.2 ± 0.2 m/s (RP) or 7.2 ± 0.2 m/s (HI); 15 min jog to finish.
Performance: 1500-m time trial on an indoor 200-m track one day before and one day after each taper; same time of day; diet repeated and no caffeine for 24 h; heats arranged so similar runners did not race each other; told to run all out without drafting or tactics; RFID splits every 100 m; no feedback until the study ended.
Analysis: two-way repeated-measures ANOVA (taper × condition) with Bonferroni post hoc; smallest meaningful change 1% (2.6 s), magnitudes as multiples of it; Wilcoxon tests for 300-m segment speeds; mean ± SD or ± 90% confidence limits.
Funding and interests: no funding from outside Loughborough University; no conflicts of interest. Authors were at the English Institute of Sport, British Athletics and Loughborough University.
Key Results
| Race-pace taper (RP) | High-intensity taper (HI) | |
|---|---|---|
| Continuous running vs baseline | 37 ± 8 km (−30%) | 21 ± 5 km (−60%) |
| Total running vs baseline | 45 ± 10 km (−32%) | 29 ± 7 km (−56%) |
| Final session speed (% of race speed) | 101 ± 1 | 110 ± 3 (prescribed 115) |
| Final session, mean 300-m time | 48.0 ± 2.1 s | 43.8 ± 2.0 s (prescribed 41.0 ± 1.5 s) |
| 1500 m before → after the taper | 258.9 ± 7.4 → 255.7 ± 8.7 s | 259.0 ± 6.3 → 253.8 ± 7.7 s |
| Improvement (± 90% confidence limits) | 3.2 ± 3.8 s (−0.6 to 7 s), p = 0.15; 1.2% | 5.2 ± 3.7 s (1.5 to 8.9 s), p = 0.03; 2.0% |
| Multiples of the smallest meaningful change (2.6 s) | 1.3 | 2.0 |
| Runners beyond the smallest meaningful change | 5 of 8 (6 faster, range 2.4-8.8 s; 2 slower by 4.7 and 5.8 s) | 7 of 8 (range 3.8-11.7 s; 1 slower by 7.0 s) |
Pacing: after both tapers the first and second 300-m segments were faster than before the taper (RP 50.4 ± 1.1 vs 52.3 ± 2.1 s and 50.1 ± 1.7 vs 52.1 ± 0.9 s; HI 49.5 ± 2.3 vs 51.8 ± 2.3 s and 49.9 ± 1.5 vs 52.5 ± 1.0 s; p = 0.012-0.017); the last three segments were run at a pace similar to before.
The runner who got worse after both tapers ran a negative split before the tapers and went out too fast after them; he had the highest training volume (84 km a week) and so the largest absolute cut.
Every runner ran the HI final session slower than prescribed (1.1 to 4.5 s per 300 m).
Where the paper does not add up:
A taper × condition ANOVA is described, but no comparison between RP and HI is reported; "a larger improvement" rests on HI being significant against its own baseline and RP not. The two confidence intervals overlap almost entirely.
The abstract and conclusion describe the final session as run "at 110% of race pace"; the design prescribed 115%, and 110% is what the runners achieved.
Table 3 marks HI's 2.0 multiples of the smallest change with one asterisk ("small") and RP's 1.3 with none, though the footnote defines one asterisk as more than 1 multiple.
The accepted manuscript's abstract ends "results in a larger improvement ... than a higher volume taper"; the published abstract instead recommends running faster than race pace late in a low-volume taper. The typeset version was not read.
Mechanisms
Measured: no physiology was measured (the authors' limitation); only performance, splits and training load.
Authors' interpretation: the larger volume cut allowed more recovery from the preceding training, so the runners could respond to a faster final session; runners may have run slower than 115% to protect themselves from fatigue so close to the trial, or wind on the outdoor track may have slowed them; after a taper runners sense their better state and start faster, which can cost an inexperienced or over-optimistic runner the race.
Inference: HI changed two things at once (continuous volume −60% instead of −30%, and the final session at 110% instead of race pace), so the study cannot say which mattered; interval volume and session frequency were the same in both.
Practical Application
What should my last hard session before a race be? In 8 trained 1500-m runners, 5 × 300 m at about 110% of race pace 3 days out, with continuous running cut 60%, went with a 5 s faster 1500 m; no direct test against race pace.
Both tapers worked for most runners; the elite-practice taper (continuous running −30%, last session at race pace) improved 6 of 8 by 2.4-8.8 s. A larger volume cut with a short, faster last session did at least as well — no evidence of a difference between them in this trial.
Day of the last session: here it was day −3 (5 × 300 m, 90 s recovery, 1,500 m of work).The base's house schedule for 4-6 h events puts one race-intensity session on days −7 to −5 and openers on day −4 (SYNTH-34, long-events section, practice); its general race-week row puts one short session at race pace or a little faster 2-4 days out and short openers the day before (practice).This trial supports a short session faster than race pace three days out for 1500-m runners (B, one trial); it does not test long events.
Rehearse racing tapered: after both tapers the runners went out faster; one runner started too fast and finished slower both times. The authors advise practising pacing in a tapered state before an important race.
Try both before it matters (the authors' advice): responses varied, so test a taper before a minor race.
Limitations
Small and unblinded. Eight men; runners knew their load had changed.
No comparison between the tapers is reported. The headline contrast is between two within-condition tests; with 90% confidence limits of ±3.7 and ±3.8 s, the 2-s difference between the means is not shown to be real.
Two variables changed at once. Continuous volume and final-session speed differed together, so their separate effects are unknown.
The intended session was not delivered. 115% of race speed was prescribed, 110% was run; outdoor conditions were not measured.
Time trial in heats, not a race. Self-selected pacing, which the authors say may have confounded the taper effect; the test was in the winter training season, not at a peak.
Accepted manuscript. Numbers were read from the author version in the university repository; the typeset article was not checked.
Funding and authorship: no external funding; the authors were at the English Institute of Sport, British Athletics and Loughborough University, and the race-pace prescription is their own group's model (REF-818).
How this entry was checked and changed
Why this level
crossover with n = 8 (under 15), unblinded to training load; read from the full text of the accepted manuscript (Nottingham Trent University repository), not the typeset version of record
Full text (checked 2026-10-26)
Performance time was improved after HI by 5.2 ± 3.7 s (mean ± 90% confidence limits, p = 0.03) and by 3.2 ± 3.8 s after RP (p = 0.15) (full text)
Pre-taper 1,500 m times were 258.9 s ± 7.4 s (RP) and 259.0 ± 6.3 s (HI), p = 1.00; post-taper 255.7 ± 8.7 and 253.8 ± 7.7 s (full text)
Continuous running was 54 ± 14 km at baseline, 37 ± 8 km in RP (-30%) and 21 ± 5 km in HI (-60%) (full text)
Mean 300 m repetition time in the final session was 43.8 ± 2.0 s in HI vs 48.0 ± 2.1 s in RP, and slower in HI than prescribed (41.0 ± 1.5 s); its mean intensity was in fact 110% of race-speed (full text)
In the HI condition, seven of eight participants improved in excess of the SMC (range 3.8-11.7 s) and one worsened (7.0 s); in RP two worsened (4.7 s and 5.8 s) (full text)
The runner who worsened in both conditions had the highest training volume before tapering (84 km per week) (full text)
The authors declare no conflicts of interest and the study did not receive funding from sources external to Loughborough University (full text)