Background Growing cross‐sectional research shows that the theory of planned behaviour (TPB) is robust in predicting intentions to use condoms and condom use behaviour. Yet, little is known about the TPB's utility in explaining intentions to use condoms and condom use behaviour over time. Methods This study used a longitudinal design and latent variable structural equation modelling to test the longitudinal relationships postulated by the TPB. School‐going youths in Ghana provided data on attitudes, subjective norms, perceived control, intentions, and behaviour regarding condom use at three time points, spaced approximately three months apart. Results As predicted by the TPB, the results showed that attitudes were significantly positively associated with intentions to use condoms over time. Contrary to the TPB, subjective norms were not significantly associated with intentions to use condoms over time. Perceived control did not predict intentions to use condoms over time. Moreover, intentions to use condoms were not significantly associated with self‐reported condom use over time. Conclusion These results suggest that school‐going youths in Ghana may benefit from sex education programmes that focus on within‐subject attitude formation and activation. The theoretical and methodological implications of these results are discussed.