Recent investigations have suggested that caffeine consumption is related to the occurrence and severity of premenstrual symptoms. Phillis has proposed that not only the total amount of caffeine consumed but also the pattern of consumption over the menstrual cycle may be important. This study explored whether women who experience moderate or severe premenstrual symptoms differ from other women in their pattern of caffeine intake throughout the menstrual cycle. Analysis of data for 96 complete menstrual cycles from 47 women demonstrated that caffeine intake during the menstrual cycle differed between women who experience moderate or severe premenstrual symptoms and other women. Furthermore, the monthly pattern of caffeine consumption for women with moderate or severe premenstrual symptoms, but not for other women, differed substantially from Phillis's proposed beneficial pattern. Women with premenstrual symptoms may self-mediate with caffeine in response to premenstrual symptoms, thereby exacerbating their symptoms.