Circadian rhythms in the Syrian hamster can be markedly phase shifted by 3 h of wheel running or arousal stimulation during their usual daily rest period ("subjective day"). Continuous wheel running is predictive but not necessary for phase shifts of this "nonphotic" type; hamsters aroused by gentle handling without running can also show maximal shifts. By contrast, physical restraint, a standard stress procedure and thus presumably arousing, is ineffective. To resolve this apparent paradox, phase-shifting effects of 3-h sessions of restraint or other stress procedures were assessed. In a preliminary study, phase shifts to arousal by gentle handling were significantly potentiated by the cortisol synthesis inhibitor metyrapone, suggesting that stress-related cortisol release may inhibit phase shifts to arousal. Next, it was confirmed that restraint in the subjective day does not induce phase shifts, but behavioral observations revealed that it also does not sustain arousal. Restraint combined with noxious compressed air blasts did sustain arousal and induced a significant cortisol response compared with arousal by gentle handling but did not induce shifts. Restraint combined with continuous horizontal rotation was also ineffective, as was EEG-validated arousal via confinement to a pedestal over water. However, 3 h of resident-intruder interactions (an intense psychosocial stress) or exposure to an open field (a mild stress) did induce large shifts that were positively correlated with indexes of forward locomotion. The results indicate that large phase shifts associated with arousal in the usual sleep period are neither induced nor prevented by stress per se, but are dependent on the expression of at least low levels of locomotor activity. Sustained arousal alone is not sufficient.