The objective of this study was to investigate whether semen quality has changed during the years 1977-1995 in a group of unselected semen donor candidates, and to determine whether semen quality is subject to seasonal variation, by analysis of time- and season-related changes in semen quality using multiple regression and ANOVA. The study was based on analysis of the first semen sample delivered by 1927 semen donor candidates in Copenhagen during the period 1977-1995, with determination of semen volume, sperm concentration, total sperm count, percentage motile spermatozoa, and a semiquantitative sperm motility score. Multiple linear regression analysis with year, sexual abstinence and season as covariates showed a significant increase in mean sperm concentration from 53.0 x 10(6)/mL in 1977 to 72.7 x 10(6)/mL in 1995 (p < 0.0001) and in mean total sperm count from 166.0 x 10(6) to 227.6 x 10(6) (p < 0.0001). Mean semen volume and percentage motile spermatozoa did not change. Sperm motility deteriorated, as the spermatozoa in 74.2% of the samples were of excellent motility in 1977-1980 compared to only 41.9% in 1993-1995 (chi 2 = 130.0, p < 0.0001). Analysis of variance showed significant variation between seasons regarding sperm concentration (p < 0.0001) and total sperm count (p < 0.0001). Highest sperm counts were found in spring, with a mean concentration (95% C.I.) of 77.6 x 10(6)/mL (71.9-83.7), and lowest in summer, with a mean of 57.5 x 10(6)/mL (50.1-65.4). No other semen parameter varied with season. It is concluded that sperm counts increased, whereas sperm motility decreased, in a group of Danish semen donor candidates, from 1977 to 1995. Due to the retrospective design and the anonymity of the donors, we were unable to control for variation in donor age, and we cannot exclude the possibility that some donor candidates were selected by being accepted as donors by other semen donor services in Copenhagen. With these limitations in mind, we suggest our results should be interpreted cautiously and regarded as a contribution to the ongoing dispute on whether or not there is a continuous decrease in sperm quality. The seasonal variations found in sperm concentration and total sperm count were pronounced and were not attributable to seasonal differences in the length of sexual abstinence. Additionally, the same seasonal pattern was observed in five successive year-intervals. These findings strongly indicate that human testicular function is influenced by season, a phenomenon well known in many lower mammals.