The effectiveness of water-training in maintaining cardiorespiratory endurance was investigated in 16 cross country athletes, 18--24 years. Following a competitive season, subjects were stress-tested (T1) and divided into three equated groups based on VO2 max. Group I (n=5) continued training as it had during the competition season. Group II (n=5) underwent an experimental period of water-training, and Group III (n=6) let their training lapse. Subjects in the water-training group exercised in deep water for 40 minutes, 6 days/week for 3 weeks, supported by a flotation device which permitted them to engage in a running type activity, resembling their natural running form. All subjects were retested after 3 weeks (T2). A non-significant F ratio from an analysis of variance at T1 confirmed the equality of the three groups in terms of VO2 max. Analysis of covariance at T2 using T1 VO2 max values as covariates revealed a significant (p less than .05) F ratio reflecting a significant (p less than .05) difference between the regular training group and the group which let its training lapse. The water-training group did not differ significantly from the regular training group indicating that the water-training programme prevented a significant decline in VO2 max.