Agreement rates for waking and sleeping obtained via sleep diary and accelerometer were evaluated, to compare the two methods. Sleep/wake data for consecutive days and nights were surveyed in 119 healthy university students. Accelerometer sleep/wake judgments obeyed the standard algorithm. Agreement rates for waking and sleeping according to accelerometer versus sleep diary, respectively, were calculated. Sleep diary data were set as a baseline. Seventy-six subjects (63.9%), 22 to 32 years of age, presented perfect data for the analysis. The mean sleep times, in minutes, judged by sleep diary and by accelerometer, were 482.3 and 629.6, respectively. The mean percentages and standard deviations of agreement on wake and sleep were 77.5% (SD = 10.2) and 86.1% (SD = 6.2), respectively. There was a significant negative relationship between the agreement rates for wake and sleep (r = -.482, p < .01). The accelerometer showed some measurement failure during waking, presumably because of the decrease in body movement. Sleep diary data during daytime appear to be more valid for detecting a sleep/wake cycle than are accelerometer data. In contrast, nocturnal sleep diary data might be supplemented by the use of an accelerometer as long as participants do not have insomnia.