The cumulative intake curves of 50 obese and 86 normal weight women during test meals were related to body mass index, age and the three factors of Stunkard & Messick's questionnaire (Journal of Psychosomatic Research, 29, 71-83), cognitive restraint, disinhibition and perceived hunger. Eating behaviour was monitored by a concealed camera and rate of intake via an electronic balance built into the table under the plate. The eating behaviour of each subject was very consistent over a series of three or six lunches of the same solid food consumed solitarily in a constant environment, with marked differences between subjects. Differences in body mass index over the whole sample were not related to the shape of the cumulative intake curve during test meals. In multiple regression analysis, the normal-weight subjects who scored high on disinhibition of restraint in response to emotions and external influences showed a more nearly constant rate of intake. In a separate multiple regression, the overweight subjects with the same characteristic showed a decelerating rate of intake. Susceptibility to hunger had no discernible relationship to the shape of the cumulative intake curve in either normal or overweight subjects. We conclude that the shape of the cumulative intake curve can be attributed more to cognitive than to biological factors.