In higher eukaryotes, circadian behaviour patterns have been dissected at the molecular level in Drosophila and, more recently, in the mouse. Considerable progress has been made in identifying some of the molecular components of the clock in the fly, where two genes, period (per) and timeless (tim), are essential for behavioural rhythmicity. The PER and TIM proteins show circadian cycles in abundance, and are part of a negative feedback loop with their own mRNAs. Within the pacemaker neurons, the PER and TIM products are believed to form a complex which allows them to translocate to the nucleus, but how they repress their own transcription is unclear. TIM is rapidly degraded by light, a feature which permits a compelling molecular description of both behavioural light entrainment and phase responses to light pulses. The regulation of per and tim is altered in different Drosophila tissues, however, and comparative analyses of the two genes outside the Diptera reveals further unusual patterns of tissue-specific regulation. Evolution appears to have modified the way in which the two genes are utilised to generate circadian phenotypes. More recently, the cloning of mouse clock genes, including putative per homologues, opens up exciting possibilities for mammalian molecular chronobiology.