This paper details the results of a meta-analysis incorporating all the appropriately augmented TPB studies in order to statistically determine the additive effects of anticipated regret (AR) both to the prediction of intentions after the TPB variables and to the direct impacts on behaviour. Over a number of studies there was a strong AR-intention relationship (r(+)=.47, k=25, N=11,254), and AR significantly and independently added to the prediction of intentions over and above the TPB variables; there was a moderate relationship between AR and behaviour (r(+)=.28, k=8, N=2,035) with AR having a direct and significant impact on prospective behaviour, and there was support for the unique contribution of AR even when accounting for attitude. Implications and issues for further research are discussed.