Eight army recruits were studied at the start, middle, and end of their initial 18-week training programme. At each point the subjects were studied for four consecutive nights in the sleep laboratory. Their sleep was characterized by the means of the recordings on the last two nights. Within 2 days of the sleep recordings (but never on the same day) each subject spent 2 non-consecutive days in the exercise laboratory. On the 1st day a maximum oxygen consumption (VO2 max) measurement was performed on a treadmill and on the 2nd day a 24-min progressive exercise bicycle ergometer test was carried out with simultaneous venous sampling (for lactic acid measurements) and oxygen consumption recordings from which the lactate turn point (LTP) was calculated. LTP was used as a measure of fitness. Approximately 1 week after the above measures lean muscle mass as calculated by total body potassium estimation was obtained for each subject. Slow wave sleep (SWS) as a percentage of total sleep time increased significantly between the start and the measurements at 9 and 18 weeks, being 21.9%, 29.9%, and 28.5% respectively. Anaerobic threshold increased significantly (P less than 0.05) over the first 9 weeks and continued to increase to the end of the training period (P less than 0.001) using VO2 when lactate level was 2 mmol/l as a percentage of VO2 max. With increase in fitness, sleep onset latency and wake time during sleep decreased and sleep efficiency improved. The results suggest that as fitness increases sleep quality improves.