Disordered gambling is best conceptualized as a continuum of severity. Previous research has demonstrated the utility of studying individuals at all points of this spectrum. The sequence of the development of gambling problems and change in gambling involvement along this continuum of severity is not well understood. The present study examined the interplay between cognitive distortions and gambling involvement in a population sample recruited in Alberta, Canada. Data from 1372 participants over 4 assessment waves (5 years) were used to generate a 2-factor latent structure using gambling fallacies and gambling involvement measurements. Structural equation modelling showed that cognitive distortions more strongly predicted future gambling involvement than the reverse relationship, using the comparative fit index (CFI) and the root mean square error of approximation (RMSEA) to assess the models. In addition, cognitive distortions declined over time, whereas gambling involvement remained stable. The results of the study suggest that focusing primarily on cognitive mechanisms in public health initiatives for gambling disorders may be a more effective strategy than focusing on behavioural solutions.