Plasma progesterone, LH and prolactin concentrations were measured twice a week in mature sows kept outdoors without boars in two experiments to examine whether perception of daylength change underlies seasonal infertility in domestic pigs. In Expt 1, melatonin implants inserted on 12 April or 22 May to block perception of the increasing daylength did not affect the oestrous cycle significantly, since only two untreated control sows became seasonally anoestrous. In Expt 2, all control sows became anoestrous for 131 +/- 42.5 days (n = 11). Melatonin implants inserted at the spring equinox (n = 9) prevented seasonal anoestrus (P < 0.001), but timed daily oral melatonin administration was less effective (P < 0.05): 5 of 11 sows became anoestrous for 132 +/- 45.6 days. In both experiments, there were significant low-amplitude seasonal rhythms in mean plasma prolactin and LH concentrations. Prolactin reached maximum concentrations 2-4 weeks before the summer solstice, whereas LH reached a nadir 4-6 weeks after the summer solstice. Neither anoestrus nor melatonin administration altered these patterns consistently. Endogenous plasma melatonin was higher (P < 0.001) during darkness (12.5 ng l(-1)) than during daytime (8.9 ng l(-1)) in untreated sows, but only clearly so during spring and summer. Melatonin implants increased mean daily plasma melatonin to 146 ng l(-1), whereas melatonin fed at 15:00 h increased values to 40-60 ng l(-1) 2-4 h after administration, but daytime concentrations were unchanged. Melatonin administration, despite decreasing seasonal anoestrus, did not prevent the seasonal increase in plasma prolactin and had no significant effect on plasma LH; therefore, its role in regulating seasonal changes in the reproduction of domestic sows remains uncertain.