The currently assumed aetiology for delayed sleep phase disorder (DSPD) is a delay of the circadian system. Clinicians have sought to use bright light therapy, exogenous melatonin or chronotherapy to correct the disorder. However, these treatments have achieved unreliable outcomes for DSPD patients and, as such, one suggestion has been that the disorder may be caused by a longer than normal circadian rhythm period length (i.e. tau). The present study investigated this premise using a 78-h ultradian, ultra-short sleep-wake cycle. This constant bedrest routine was used to simulate a series of 1-h long 'days' by alternating 20-min sleep opportunities and 40 min of enforced wakefulness. Thirteen participants were recruited for the study including, six people diagnosed with DSPD according to the International Classification of Sleep Disorders-2 [mean age = 22.0, standard deviation (SD) = 3.3] and seven good sleepers (mean age = 23.1, SD = 3.9) with normal sleep timing. The DSPD participants' core temperature rhythm tau (mean = 24 h 54 min, SD = 23 min) was significantly longer (t = -2.33, P = 0.04, Cohen's d = 1.91) than the good sleepers' (mean 24 h 29 min, SD = 16 min). The temperature rhythm of the DSPD participants delayed more rapidly (i.e. >25 min day(-1) ) than the good sleepers'. These findings provide an explanation for the difficulty that DSPD patients have in phase advancing to a more conventional sleep time and their frequent relapse following treatment. The outcomes of this study support a vigorous and continued application of chronobiological and behavioural therapies to entrain DSPD patients to their desired earlier sleep times.