Controversial claims, based on a meta-analysis aggregating 61 heterogeneous observational studies, have been made that human sperm output has decreased by 50% over the last six decades and that this trend may be due to global pollution. If true, such effects should be evident in all areas of the globe; however, longitudinal studies within single centres in Europe and America have produced conflicting results and there are no reports from the southern hemisphere. We therefore reviewed semen analyses obtained from 1980-1995 from 689 healthy men volunteering for screening either as potential sperm donors for a donor insemination programme (n = 509) or to participate in five male contraception research studies (studies no. 1-5, n = 180). All were recruited through the Andrology Unit of the Royal Prince Alfred Hospital, Sydney, by the same doctors using standard methods of recruiting, screening and laboratory examination throughout the period 1980-1995. Recruitment was by advertising without regard to marital or fertility status except in two contraceptive efficacy studies (no. 1 and no. 3) where participants had to be in a stable relationship requiring contraception. Analysing the first semen sample individually or when grouped by year of ejaculation, there was no significant difference in sperm concentration over time or between years or according to year of birth. During the second half of this period, 180 consecutive volunteers were recruited by the same doctors and staff for five male contraception studies. The median sperm concentration for studies no. 1 (103 x 10(6) ml) and no. 2 (142 x 10(6) ml) were significantly (P < 0.05) higher than for studies no. 3-5 (84, 67 and 63 x 10(6) ml, respectively) and for potential sperm donors (median 69 x 10(6) ml). The inconsistency of these estimates illustrates the magnitude of bias (up to 100%) in sperm output that may occur in recruiting groups of self-referred volunteers within a single centre. This highlights the invalidity of extrapolating similar findings on sperm output of self-selected volunteers to the general male community or in using such study groups to characterize sperm output in supposedly 'normal' men.