This paper analyzes whether stock return dispersion (cross-sectional variance of equity portfolio returns) provides useful information about future stock returns, both at the aggregate and portfolio levels. Return dispersion consistently forecasts a decline in the (excess) stock market return, and compares favorably with alternative predictors. Furthermore, return dispersion outperforms the alternative variables in forecasting the (excess) market return out-of-sample. The results from both in-sample and out-of-sample regressions show that return dispersion has greater forecasting power for large and growth stocks compared to small and value stocks, respectively. Return dispersion also helps to forecast stock market volatility at short horizons.