In this study of sell-side analysts’ forecasts, we explore the effects of analyst aptitude, learning-by-doing, and the internal environment of the brokerage house on forecast accuracy. Our results indicate that analysts’ aptitude and brokerage house characteristics are associated with forecast accuracy, while learning-by-doing is only associated with forecast accuracy when we do not control for analysts’ company-specific aptitude in forecasting. It is unlikely that this result is caused by measurement errors because it is robust when we use a sub-sample where we can accurately measure experience.