The following is a progress report of our studies to identify important features of the circadian pattern of melatonin secretion which provide the photoperiodic code for daylength in regulating seasonal breeding in the Suffolk ewe. The first series of experiments evaluated two conceptual models of how melatonin codes for daylength: the circadian timing of the melatonin elevation as opposed to the length of the time melatonin is elevated during each 24-hr period (phase vs duration). Strong support has been gathered for the duration hypothesis. No evidence was obtained to support a role for phase; nevertheless, this hypothesis could not be discounted definitively. A second series of studies evaluated the importance of the previous melatonin pattern to the interpretation of a given melatonin signal. Evidence is presented that a fixed melatonin pattern can maintain a given reproductive response only for a limited length of time and that this response can be prolonged by appropriate changes in the melatonin pattern. Thus, change is an important feature of the melatonin signal. Further, the nature of the melatonin change appears to be crucial, specifically whether the nocturnal elevation increases or decreases in duration. Thus, transfer to a common photoperiod can promote either reproductive induction or arrest, depending upon whether the transfer leads to a decrease or increase in daylength. This has important ramifications to the photoperiodic timekeeping process in those species of mammals which utilize daylength to time their seasonal reproductive cycle.