This paper concerns methodological limitations in research on nonphotic res
etting of circadian rhythms. There are problems in producing phase response
s curves for arousing activity-inducing stimuli when locomotor activity is
also used as a phase marker. It is also difficult to define the nature of t
hese nonphotic inputs. Dose-response curves relating amount of wheel runnin
g to phase shifts have been overinterpreted, and the measurement of phase s
hifts is complicated by concomitant changes in period. Some of these points
also apply sometimes to chronobiological experiments with light.