Pownall et al. (1993) document that nearly 80 percent of their sample of voluntary management earnings forecasts are not precise point forecasts. Imprecise forecast forms include closed-interval forecasts (i.e., ranges), open-interval forecasts (i.e., minimums and maximums), and general impressions about firms' earnings prospects. We perform cross-sectional logistic regressions to document determinants of forecast precision. Our sample consists of 1,212 annual and interim management forecasts. After controlling for firm-specific and horizon- specific earnings uncertainty, we find that managers produce more precise forecasts of annual earnings for firms with greater analyst following (our proxy for private information) and for smaller firms (our proxy for public information). The results are robust across subsamples. The majority of the results, however, do not hold for interim forecasts.