Drawing on the alliance learning literature, we propose that a firm's relative capability to learn partner-specific know-how holds the key to understand the learning race phenomenon and its performance consequences. Specifically, we argue that a firm with higher specific learning capability relative to its partner's will be rewarded with superior stock performance. We also contend that equity alliance governance and market similarity between partners moderate this relationship in opposite directions. Equity alliance governance motivates firms to suppress competitive learning and thus reduces the positive impact of the specific learning capability gap on abnormal stock returns, while market similarity between partners aggravates the learning race and strengthens the positive impact of the specific learning capability gap. Results of an event study on 610 R&D alliances in the US computing and biopharmaceutical industries in the period of 1984-2003 support our hypotheses. Our study contributes to a better understanding of the learning race, its contingencies, and its performance implications.