The time in spring when a male kestrel rapidly increases his daily hunting time and his hunting yield, and thereby the amount of food delivered to the female, determines the date when she lays the first egg. Food experiments in free-living and captive kestrels gave a significant advance in laying date. Clutch size, which decreases with progressive laying date, did not change independent of date in response to food manipulation. These effects are in agreement with most other feeding experiments. Photoperiod experiments in kestrels advanced the reproductive cycle in constant long days, and a similar seasonal decline in clutch size was found. It seems that there is an internally preprogrammed decrease in clutch size within an annual "reproductive window". A proximate control model for the seasonal decline of clutch size is proposed, modified from an earlier model by HAFTORN (1985). This incorporates an increasing tendency to incubate the first eggs with progression of the season, an egg contact-incubation positive feedback loop, and the resorption of further follicles in the ovary when the laying female incubates 50% of the time. This follicle resorption fixes the clutch size ca. four days before the last egg is laid. the 50% incubation level is reached earlier in late females and consequently resorption starts earlier and the resulting clutch is smaller than in early females. Experiments in kestrels with removal and addition of eggs, in combination with measurements of incubation behaviour are discussed in relation to the model. Plasma prolactin data of female kestrels show that this hormone is a serious candidate for a physiological component relaying time of year in our model for clutch size regulation.