Consider a variable y (e.g., first year GPA at a given college) whose regression line on a predictor variable x (e.g., high school grade average) varies over a fairly large number of populations (high schools) which slopes and intercepts of the regression lines may be estimated under both school-to-school and year-to-year variation, and a method is indicated for forecasting the slope and intercept for one year beyond the data set (i.e., for the current year). The predictive distribution of this slope and intercept may be estimated, yielding a prediction interval for the y -value of any student in the current year, given his or her slope and x -value. (l6pp.)