In a residential research ward the reinforcing and subjective effects of caffeine were studied under double-blind conditions in volunteer subjects with histories of heavy coffee drinking. In Experiment 1, 6 subjects had 13 opportunities each day to self-administer either a caffeine (100 mg) or a placebo capsule for periods of 14 to 61 days. All subjects developed a clear preference for caffeine, with intake of caffeine becoming relatively stable after preference had been attained. Preference for caffeine was demonstrated whether or not preference testing was preceded by a period of 10 to 37 days of caffeine abstinence, suggesting that a recent history of heavy caffeine intake (tolerance/dependence) was not a necessary condition for caffeine to function as a reinforcer. In Experiment 2, 6 subjects had 10 opportunities each day to self-administer a cup of coffee or (on different days) a capsule, dependent upon completing a work requirement that progressively increased and then decreased over days. Each day, one of four conditions was studied: caffeinated coffee (100 mg/cup), decaffeinated coffee, caffeine capsules (100 mg/capsule), or placebo capsules. Caffeinated coffee maintained the most self-administration, significantly higher than decaffeinated coffee and placebo capsules but not different from caffeine capsules. Both decaffeinated coffee and caffeine capsules were significantly higher than placebo capsules but not different from each other. In both experiments, subject ratings of "linking" of coffee or capsules covaried with the self-administration measures. These experiments provide the clearest demonstrations to date of the reinforcing effects of caffeine in capsules and in coffee.