A group of randomly selected males, over 50 years of age, was examined with regard to history and urinary flow rate. Originally, the group had comprised 93 men without voiding problems, 15 who, although suffering from voiding problems, had not yet contacted their doctor, and 4 who had contacted their doctor because of voiding problems. Five years later the data were re-examined and 61 men were fully investigated again; in the remaining 51 patients only the history was updated. The 2 sets of data were evaluated separately as 2 cross-sectional investigations and as paired data sets by means of non-parametric statistical analysis. All uroflow variables were considered, i.e. Qmax, Qave, Qmax-time, Q "corrected", volume and the ratio Qmax/Qmax-time. In addition, the flow curve configuration was classified as normal or abnormal. When compared as 2 cross-sectional investigations, the latter confirmed the results of the first. Qmax, Qave, Q "corrected" and volume decreased with advancing age, but no correlation could be demonstrated between Qmax-time and age. The flow curve pattern altered, so that a normal flow curve was seldom seem at an advanced age. Uroflow variables tended towards the abnormal with advancing years and symptoms increased concurrently. These changes were largely accepted by the elderly males. Thus the decision to operate for benign prostatic hyperplasia should be based on both history and urinary flow.