In six healthy subjects with a history of caffeine-induced wakefulness caffeine kinetics were compared to with those in six subjects not affected by caffeine. The data indicated that the former have a longer plasma t 1/2 (mean 7.4 and 4.2 hr) and slower plasma clearance (mean 1.2 and 1.7 ml . min-1 . kg-1) of caffeine. Plasma caffeine concentration at midnight, 8 hr after afternoon coffee, is higher in those with caffeine-attributed insomnia. Those reporting "coffee wakefulness" also tend to drink less coffee. We conclude that the rate of caffeine metabolism is a determinant of individual variation in the effect of drinking coffee on sleep.