The rhythm in melatonin formation was studied in rats kept in natural daylight throughout the year and in rats housed under different artificial photoperiods. N-acetyltransferase activity was chosen as the indicator of melatonin formation. The period of the high night N-acetyltransferase was shortest in rats kept under LD 16:8 and longest in rats under LD 8:16. A short, 1-day adaptation to longer days, from LD 12:12 to LD 16:8, involved the induction of the morning decline in N-acetyltransferase by an earlier ‘light on’. Following long-term adaptation to LD 16:8, N-acetyltransferase declined spontaneously 1 h before the morning ‘light on’. A short, 1-day adaptation to shorter days, from LD 12:12 to LD 8:16, involved an earlier evening rise in N-acetyltransferase and a spontaneous morning decline at approximately the same time as in LD 12:12. Following long-term adaptation to LD 8:16, the period of high N-acetyltransferase shifted towards the later part of the night and the activity declined 2 h before the ‘light on’. In natural daylight, the period of the high night N-acetyltransferase was shortest in June and longest in December. The morning decline in N-acetyltransferase was probably induced by dawn on April 19 and June 20, but was spontaneous on December 19. The existence of differences in the N-acetyltransferase rhythm and consequently in the rhythm of melatonin formation in natural daylight throughout the year and under different artificial photoperiods supports the hypothesis that the changing rhythmic pattern of melatonin release might be a mechanism by means of which a photic information is transformed into a humoral one.