To what extent can one infer that superior capabilities are driving sustained superior performance? Modeling performance as some combination of differences in capabilities and processes of cumulative advantage, we argue that a Bayesian framework in which decision makers take into account the differences in cumulative advantage provides for a correct inference. We show, using both simulated and real performance data, that the Bayesian method gives rise to estimates relevant for the inference problem. The estimates also illustrate why a firm with superior performance during a longer period can be less likely to possess superior capabilities than a firm with superior performance during a shorter period. Our work has implications for the origins of competitive advantages and for organization learning in strategy research. Copyright © 2012 John Wiley & Sons, Ltd.