Progress in habit theory can be made by distinguishing habit from frequency of occurrence, and using independent measures for these constructs. This proposition was investigated in three studies using a longitudinal, cross-sectional and experimental design on eating, mental habits and word processing, respectively. In Study 1, snacking habit and past snacking frequency independently predicted later snacking behaviour, while controlling for the theory of planned behaviour variables. Habit fully mediated the effect of past on later behaviour. In Study 2, habitual negative self-thinking and past frequency of negative self-thoughts independently predicted self-esteem and the presence of depressive and anxiety symptoms. In Study 3, habit varied as a function of experimentally manipulated task complexity, while behavioural frequency was held constant. Taken together, while repetition is necessary for habits to develop, these studies demonstrate that habit should not be equated with frequency of occurrence, but rather should be considered as a mental construct involving features of automaticity, such as lack of awareness, difficulty to control and mental efficiency.