To estimate the reliability of task-fMRI, we conducted a repeated measure fMRI study of psychomotor vigilance task once per day and 30 days in total. Salivary cortisol and sleep duration were taken into consideration based on previous study. However, we found dramatic within-subject variability in the brain activity of dACC and primary motor cortex. The cortisol showed insignificant association with brain activity, but sleep duration affected the motor and temporal cortices. Such finding suggests that the PVT-based fMRI showed large variability across different days, which might be associated with the deviations of sleep duration.