In badminton, the short doubles serve requires accuracy, whereas the long singles serve also requires power. The present study investigated if there are time-of-day effects for these two types of serve. Eight male badminton players completed two familiarization sessions before being tested at 08:00, 14:00 and 20:00 hours, the session times being counterbalanced. At each session, subjects performed ten warm-up serves, followed by ten short and ten long test serves. A 1 m2 yellow grid was placed at the front of the recipient's service box (short serve) or at the back of the court along the centre line (long serve). Subjects scored 'hits' with their serves by putting the shuttle in the grid as close as possible to one corner of the target grids. Accuracy was assessed by calculating the average radial error (RE) of the serves, and consistency by calculating the standard deviation of the RE. Data were analysed by ANOVA. There was a time of day effect for temperature (p < 0.0005) and a trend in this for grip strength (p = 0.052). The accuracy of the 'hits' was highest at 14:00 hours (p < 0.05) and the short serve showed greater accuracy than the long serve (p < 0.0005). There was a significant effect of serve number with the short serve, subjects getting more accurate with later serves (p = 0.036). No significant variation in consistency was found (p > 0.05). In summary, the lack of agreement in profiles for temperature and the accuracies of long and short service suggests that some factor(s) that deteriorate during the course of the time spent awake might be involved, in addition to body temperature, in determining accuracy in the tasks measured.