Sleep is an important component of health, with numerous studies linking both sleep duration and sleep quality to measures of well-being, and also to metabolic measures such as insulin resistance and glycemic control. Polysomnography (PSG) done in a sleep study unit is the gold standard, but this is expensive and inconvenient. Many personal trackers give a measure of sleep, but how accurate are they, and are they accurate enough for clinical use? The answer is probably. Two separate studies appearing today in the Journal of Clinical Sleep Medicine are helpful. The first compared sleep parameters in people with type 1 diabetes, measured with PSG or an Actiwatch. The Actiwatch offers a choice of three different sensitivity thresholds, and only the low sensitivity threshold fared well. The second study compared the Motionwatch 8 with a number of validated sleep questionnaires, and found that 5 to 7 nights of Motionwatch data were needed to get reasonable correlations. Overall the reliability of these personal trackers was lower than reported by the manufacturers. Using the output of personal trackers clinically should be done carefully, with longitudinal data being the most helpful.