Objectives: To determine the predictors of Intention and distance walked in a standardised 6-Minute-Walk-Test from the Theory of Planned Behaviour (TPB) in elderly patients with congestive heart failure. Method: Stable CHF patients over 70 years old in NYHA Class II-III were recruited for a randomised controlled trial of exercise. A 12 item TPB questionnaire assessing attitude, subjective norm, perceived behavioural control and intention to "take gentle exercise for your heart condition" was administered before exercise training (T1) and again 12 weeks later (T2). A standardised 6-minute-walk-test was also carried out at T1 and T2 Results: 82 patients; mean age = 81, Female = 36 (44%); Class 11 = 46 (56%). The internal consistency of the TPB measures was generally poor at T1, but more satisfactory at T2 (Cronbach's Alpha for attitudes, subjective norms (SN), perceived behaviour control (PBC) and intention at T1 were 0.23, 0.64, 0.51 and 0.85, respectively, at T2 were 0.68, 0.74, 0.57 and 0.82). Only Subjective Norm correlated reliably with Intention at T1, but all the TPB components correlated with Intention at T2. In a multiple regression of Intention on Attitude, SN and PBC were reliable independent predictors. The proximal predictors from the TPB, Intention and PBC did not predict distance walked at T1 but did at T2. T2 correlations with distance walked (Intention = 0.22, PBC = 0.39). In a multiple regression PBC was the only reliable predictor of distance walked. Intention and PBC at T1 did not predict distance walked at T2 Conclusions: At T2, when the measures were more reliable, the components of the TBP predict Intention and PBC is a reliable predictor of distance walked. This supports the application of the TPB in this elderly disabled population.