A review of 13 studies of the direct observation of eating was undertaken in an effort to learn what these studies tell us about the causes and control of obesity. The studies assessed several measures of eating behavior but few of them assessed the same measures. There was no agreement about a distinctive 'obese eating style', but two measures showed promise in discriminating obese from non-obese persons. The first was food choice: obese persons chose more food than non-obese persons (and men chose more than women and tall persons more than short ones). The second measure was rate of eating: obese persons consumed more food per minute than non-obese persons. The studies point up the remarkable plasticity of human eating behavior and the wide range of factors which influence it. This plasticity has profound implications for both research and therapy. It means that only unusual care in the identification and control of extraneous variables will permit adequate assessment of the variables under investigation. Its second implication is as important: if eating is so dependent upon environment, it may be easier to modify than previously thought.