Pairs-trading is a short-term, self-financing arbitrage strategy in which buy and sell positions are simultaneously placed on two stocks whose prices have moved temporarily apart after following a long parallel path. We develop a new pairs-trading rule based on financial analysts’ buy/hold/sell recommendations from IBES Details Recommendation Database and test it for the period 1994–2009. On the basis of the Fama–French (1993) and Carhart [Journal of Finance 52(1), 57–82, 1997] four-factor models, we find that our trading rule generally results in positive risk-adjusted returns. It is more effective on small- and midcap pairs of stocks than on large-cap pairs, consistent with the hypothesis of information disparity in the stock market. It is more effective in the industries of mining, finance, and services than in others. In additional exploration of our strategy, we examine the correlation of analyst recommendations with past stock investment and corporate earnings performance in the past. We find significant positive correlation, lending new support to prior findings of the relation between recommendations and recent performance.