The process by which analysts revise quarterly earnings forecasts is analyzed and compared to the way in which several time-series models of quarterly earnings revise forecasts. A significant portion of the analyst's forecast revision is explained by the most recent one-quarter-ahead forecast error. Analyst revisions are adaptive in the same manner that single-period ahead model forecasts are adaptive. At longer horizons, the evidence is that analysts revise forecasts in the same way that autoregressive models of quarterly earnings revise and not as moving average models do.