To measure percent body fat (%F) using hydrodensitometry,“true”underwater weight (UWW) has conventionally been determined from the mean of 8, 9, or 10 UWW trials. Because the procedure is time‐consuming, many subjects are uncomfortable in water, and the elderly often fatigue rapidly, it would be desirable to reduce the number of trials. In 400 Caucasian men (age 35.5 years, height 176.9 cm, mass 85.7 kg) we compared %F estimates derived by four selection methods based on the mean of: (1) three trials with the highest adjacent trial correlations (trials 4, 5, 6) (%P = 22.9%); (2) three successive trials within 100 g (3 ROW) of UWW (22.9%), (3) any three trials within 100 g (3 WIN) of UWW (22.8%), and (4) trials 8, 9, and 10 (22.8%). Repeated measures analysis of variance (ANOVA) revealed no significant (p>0.05) difference between groups in %F using the four respective methods for selecting true UWW. A repeated measures ANOVA, using %F as the dependent variable, comparing selection methods across age quartiles revealed no significant (p>0.05) differences among the methods and no significant“age by trials”interaction. These data suggest that, regardless of age, in contrast to the convention of performing 10 or more trials to estimate %F, methods 3 ROW, 3 WIN, and T4‐6 are sufficient to achieve true UWW, resulting in a valid estimation of %F.