Abstract Larval cultures of the flesh-fly Sarcophaga argyrostoma maintained in circadian ‘resonance’ experiments produced a high incidence of pupal diapause when the period of the light cycle was close to ( T) 24, 48 or 72 hr, but a low incidence of diapause at T 36, 60 or 84 hr. Cultures pre-programmed for diapause by exposing pregnant females to long nights indicated the induction of non-diapause development at T 36, 60 and 84, whereas cultures pre-programmed for diapause-free development by exposing females to continuous light indicated the induction of diapause at T 24, 48 and 72. Raising the temperature reduced the heights of the diapause peaks whereas lowering the temperature raised them. With progeny from long-night-reared flies the lowest temperature tested (18°C) produced a result indistinguishable from an ‘hour-glass’ response, warning that ‘negative’ resonance experiments may merely indicate non-permissive conditions for demonstrating the involvement of circadian rhythmicity in insect photoperiodism. The results of the ‘resonance’ experiments and the effects of temperature are interpreted in terms of a multioscillator ‘external coincidence-photoperiodic counter’ model for the clock.