Temporary geographical proximity in the form of face-to-face contact is commonly thought to enhance learning. In a sample of individuals (n = 294) involved in knowledge transfer in the Olympic Games, temporary geographical proximity emerges as a rather weak predictor of learning, although its explanatory value improves when coupled with organized proximity. This association disappears, however, when controlling for other predictors, suggesting that there is no unique effect of temporary geographical proximity on learning. Part of the effect of temporary geographical proximity is mediated through other variables, urging further research into the paths of mediation. Several practical implications for knowledge transfer in mega-events result.