Asking about intentions and behaviour may alter future reports of health behaviours due to the stability of the behaviours (behavioural stability hypothesis), or changes in performance context (context stability hypothesis). Two studies (Ns = 116, 177) confirmed the distinction between stable and unstable behaviours and explored context stability for six health-related behaviours. Study 3 used a longitudinal intervention design in which the intervention group (N time 1 = 292, N time 2 = 149) reported their intentions and past-behaviours at time 1 while the non-intervention group (N = 118) did not. The context stability hypothesis was supported.