Five surveys, using a previously developed high-performance liquid chromatography procedure to measure caffeine concentrations, indicated great variations in the concentrations of caffeine in tea and coffee. In the study of beverages prepared at home, data on caffeine concentrations in 58 samples of tea and coffee, volumes of cups, and numbers of cups consumed/day, indicated that the range of caffeine intakes for the women participating was 49-1022 mg/day. There were considerable day-to-day variations in caffeine contents in coffee samples from some commercial coffee shops. When 17 samples of five national brands of instant coffee were made into beverages in the laboratory, variations in caffeine concentrations between lots were small but between brands were significant. A considerable range of caffeine concentrations was also found when 12 samples of coffee prepared at work by different individuals using the same jar of instant coffee were analysed. Analysis of tea samples prepared in the laboratory indicated that steeping time had an important influence on resulting caffeine and theobromine concentrations. People preparing their own beverages were found to drink more liquid than the volume offered commerically. The mean caffeine 'contents' of home-made coffee and of coffee prepared by individuals at work were 79.4 and 81.7 mg/cup respectively, indicating a mean intake of approximately 80 mg caffeine/cup. When this amount (80 mg/cup) was used to estimate daily intakes of caffeine from coffee, on the basis of the number of reported cups/day, and the values obtained were compared with the amounts actually consumed by individuals, the potential for misrepresentation of individual consumption became obvious. For example, for subjects consuming three cups of coffee, only 25% would have been correctly categorized in the expected range for the daily intake of caffeine, 39% would have been overestimated and 36% underestimated for the amount of caffeine consumed. These variations in caffeine concentrations and in the volume of coffee consumed have frequently been ignored in examinations of the possible relationship between coffee consumption and various health problems, and this could perhaps partly explain some conflicting results seen in epidemiological studies.