Increase in pile capacity after initial driving has been well observed in clays and sands over decades. The phenomenon is referred to as pile set-up by geotechnical engineers. More economical pile design may benefit from this time-dependent increase subject to a reliable prediction. Simple empirical relations of the current capacity with the initial capacity and elapse time after driving are available in the literature with different model parameters being suggested for clays and sands, respectively. Nevertheless, appropriateness of the relations and confidence interval of the model parameters are rarely investigated and this hinders the application of these formulae. In this study, a revised single-parameter empirical relation is proposed based on the existing formulae. A comprehensive database from pile field tests data in clayey and sandy ground in the literature is compiled and Bayesian analysis is conducted on both these empirical formulae independently for clays and sands. Bayesian inference allows not only the estimation of the uncertain parameter but also the quantification of the associated uncertainty in the form of probability distribution. This study sheds lights on the confidence interval of the model parameter and it provides designers more reliable prediction of the additional capacity due to pile set-up.