BACKGROUND: The transtheoretical model (TTM) proposes that individuals move stage and hence change behaviour through the use of processes of change (POC)--cognitive, emotional and behavioural activities related to the target behaviour. PURPOSE: This study was a test of the psychometric properties of the POC scale. METHOD: The sample consisted of year 9 students from 26 schools who completed a computer-based version of the POC questionnaire on up to three occasions. There were 1,160 former and current smokers who completed the questionnaire on the first occasion, 978 on the second, and 836 on the third occasion. Confirmatory factor analysis was used to explore three alternative factor structures. Cronbach alpha and test-retest reliability coefficients were examined. Construct validity was examined by calculating POC means for each stage and testing differences between means using analysis of variance. RESULTS: The TTM-hypothesised second-order factor that clustered ten primary factors into experiential and behavioural processes fitted the data reasonably, though an alternative model, removing the second-order factors, was a significantly improved fit. Cronbach alpha coefficients were reasonable except for two processes. Test-retest reliabilities were modest. Experiential processes showed differences across precontemplation to preparation, but differences between preparation, action, and maintenance were slight. Behavioural processes showed linear differences between stages, though helping relationships was not different across stages. CONCLUSIONS: There is evidence of some specific difficulties with the POC questionnaire that may be overcome with some reformulation. The POC questionnaire shows evidence of adequate factorial validity, sufficient to allow its use in future predictive testing.