AbstractThe aim of this study was to investigate which characteristics of the nocturnal melatonin signal, in addition to its duration, convey photoperiodic information to the reproductive axis. To achieve control over the pattern of circulating melatonin, male Syrian hamsters held under stimulatory long daylengths (16h light:8h dark) were pinealectomized to remove the principal source of circulating endogenous hormone and then fitted with chronic subcutaneous cannulae through which programmed infusions of melatonin solution or vehicle could be delivered. Experiment 1 tested whether long intervals between successive melatonin signals impaired the photoperiodic response. Animals which received a short day‐like melatonin infusion of 10 h duration once every 24 h (T = 24) for 6 weeks underwent gonadal atrophy. When the same number of signals (42) was delivered at a frequency of once every 32 h (T = 32), they were ineffective and animals remained gonadally active. Two infusion patterns were used to determine if the loss of response to 10 h signals given at T = 32 h was a consequence of the frequency per se or the long interval between signals (22 h). In the first, a ‘chimaeric’ signal which combined a long duration i.e. short day‐like 18 h melatonin signal with a short day‐like melatonin‐free interval of 14 h (combined signal T = 32 h) was able to induce significant, but only partial, gonadal atrophy. Second, when the 22‐h interval between 10‐h melatonin signals was interrupted by a short (2 h) melatonin pulse, significant but partial gonadal regression again occurred. Moreover, the response depended upon the timing of the 2 h pulse. When this fell early in the melatonin‐free interval, leaving a large portion of it intact, it had no effect on gonadal condition. In contrast, a pulse delivered in the middle of the interval, which divided it up into two short day‐like segments of 10 h each, was partially effective in restoring a short day response. The second experiment tested whether melatonin signals delivered at a high frequency would induce a photoperiodic response. A 10 h infusion delivered once every 24 h caused gonadal atrophy. The same melatonin infusion delivered at a periodicity of 20 h (T = 20) was also very potent as a short day stimulus. However, when 10‐h signals were delivered at the higher frequencies of once every 18 or 16 h, they were less effective. Only a minority of animals exhibited gonadal atrophy and overall the group means were not significantly different from those of saline‐infused controls, but were significantly greater than those of the 24 and 20 h groups. These data demonstrate that the photoperiodic response to the melatonin signal is sensitive to the frequency at which the signal is received. However, there is no evidence for a circadian basis to this sensitivity, nor a dependence upon the relationship between the endocrine stimulus and the light‐dark cycle, insofar as signals encountered at a non‐circadian period of 20 h are very effective. Moreover, the effectiveness of signals encountered at longer periodicities can be modified by manipulation of the uninterrupted duration of the interval free of melatonin, demonstrating a role in photoperiodic time measurement for the duration of t