This paper provides descriptive data on standard sources of analyst forecasts used in accounting research: the Value Line Investment Survey, the Institutional Brokers Estimate System (IBES), and, toa lesser extent, the Standard & Poor's Earnings Forecaster and Zacks Investment Research. We examine the relative accuracy of seven forecast error metrics, using various combinations of Value Line and IBES forecasts of quarterly earnings per share (EPS) and actual earnings as reported by Value Line, IBES, and Compustat. (Appendix A reports the relative accuracy of a forecast error metric based on a smaller sample of Standard and Poor's forecasts of annual EPS). We find the forecast error metric that pairs the Value Line forecast EPS with the Value Line actual EPS produces the smallest absolute forecast errors. We also test the association of these forecast error metrics with threeday excess returns centered on the data of a quarterly earnings announcement. The strongest associations are obtained with the use of Value Line actual earnings and either Value Line or IBES forecast data. This suggests that the choice of actual EPS data is more crucial than the source of forecast EPS data. Our overall conclusion is that Value Line