Spinal shrinkage was used as an indicant of loading on the spine in circuit weight-training and running regimes. The loss of stature during two sets of a circuit of weight-training (n = 10), a 6 km run by novices (n = 9) and a 25 km run by trained runners (n = 7) was assessed in male subjects. Shrinkage was not significantly different between the weight-training regime and the 6 km run by novices, mean losses being 5.4 and 3.25 mm respectively. The rate of height loss in the experienced runners was 2.35 mm over 6 km run at 12.2 km.h-1, representing 0.4 mm.km-1 over the 6 km run, this shrinkage rate being continued over the last 19 km run at 14.7 km.h-1. The loss of height could not be predicted from a set of covariates. The magnitude of the circadian variation, mean 14.4 mm, exceeded the change in height during the 25 km run. The diurnal variation conformed to a cosine function, though a better fit was obtained with a power function equation. A marked diurnal pattern was also observed in lumbar extension. Though reversal of spinal shrinkage was observed during a night's sleep, no significant recovery occurred during a 20 min resting period immediately following the exercise regimes. These results have implications for the warm-up and timing of exercise regimes that impose significant loading on the spine.