The relationship between habitual coffee and tea consumption and cognitive performance was examined using data from a cross-sectional survey of a representative sample of 9003 British adults (the Health and Lifestyle Survey). Subjects completed tests of simple reaction time, choice reaction time, incidental verbal memory, and visuo-spatial reasoning, in addition to providing self-reports of usual coffee and tea intake. After controlling extensively for potential confounding variables, a dose-response trend to improved performance with higher levels of coffee consumption was observed for all four tests (P < 0.001 in each case). Similar but weaker associations were found for tea consumption, which were significant for simple reaction time (P = 0.02) and visuo-spatial reasoning (P = 0.013). Estimated overall caffeine consumption showed a dose-response relationship to improved cognitive performance (P < 0.001 for each cognitive test, after controlling for confounders). Older people appeared to be more susceptible to the performance-improving effects of caffeine than were younger. The results suggest that tolerance to the performance-enhancing effects of caffeine, if it occurs at all, is incomplete.