Purpose: To describe pacing strategy in a 24-h running race and its interaction with sex, age group, athletes' performance group and race edition. Methods: Data from 398 male and 103 female participants of 5 editions were obtained based on a minimum 19.2-h effective-running cut-off. Mean running speed from each hour was normalised to the 24-h mean speed for analyses. Results: Mean overall performance was 135.6 ± 33.0 km with a mean effectiverunning time of 22.4 ± 1.3 h. Overall data showed a reverse J-shaped pacing strategy, with a significant reduction in speed from the second last to the last hour. Two-way mixed ANOVAs showed significant interactions between racing time and both athletes' performance group (F = 7.01; P < 0.001; p 2 = 0.04) and race edition (F = 3.01; P < 0.001; p 2 = 0.02), but not between racing time and both sex (F = 1.57; P = 0.058; p 2 < 0.01) and age group (F = 1.25; P = 0.053; p 2 = 0.01). Pearson's product-moment correlations showed an inverse moderate association between performance and normalised mean running speed in the first 2 h (r = -0.58; P < 0.001) but not in the last 2 h (r = 0.03; P = 0.480). Conclusions:While the general behaviour represents a rough, reverse J-shaped pattern, fastest runners start at lower relative intensities and display a more even pacing strategy than slower runners. The 'herd behaviour' seems to interfere with pacing strategy across editions, but not sex or age group of runners.