The effects of moderate workload and 72 h of sleep deprivation were studied using a modified continuous-performance paradigm. Ten subjects were tested hourly on a number of perceptual and cognitive tasks designed to require approximately 30 min to complete, with the remainder of each hour free. As sleep deprivation continued, the average time on task increased at an accelerating rate. The rate of increase differed among tasks, with longer tasks showing greater absolute and relative increases than shorter ones. Such increases confound sleep deprivation and workload effects. In this paper, we compare the advantages and disadvantages of several experimental paradigms; describe details of the present design; and discuss methodological problems associated with separating the interactions of sleep deprivation, workload, and circadian variation with performance.