APPROACHES TO SHORT-RANGE PROJECTION INCLUDE THE FOLLOWING: BUSINESS AS USUAL (BAU), WHICH INVOLVES A SIMPLE LINEAR EXTRAPOLATION, BASED ON THE ASSUMPTION THAT THE NEXT YEAR WILL BE THE SAME AS THIS YEAR; INPUT/OUTPUT ANALYSIS, INVOLVING SEPARATE ANALYSES OF FACTORS INFLUENCING INPUT (INTAKE FROM COURTS AND PAROLE VIOLATIONS) AND OUTPUT (PAROLE AND DISCHARGE); AND MULTIPLE REGRESSION ANALYSIS, A LINEAR EXTRAPOLATION METHOD THAT OFFERS CERTAIN IMPROVEMENTS OVER THE BAU TECHNIQUE. EXPERIENCE SUGGESTS THAT THERE IS A GREATER NEED TO DEVELOP METHODS THAT GIVE EARLY WARNING OF MAJOR CYCLICAL CHANGES RATHER THAN REFINE METHODS OF SHORT-TERM PREDICTION. A METHOD FOR MAKING INTERMEDIATE FORECASTS CAN BE ADAPTED FROM TECHNIQUES USED IN TECHNOLOGICAL FORECASTING. THE METHOD IS A HEURISTIC TECHNIQUE RATHER THAN A STATISTICAL ONE, PROVIDING INFORMATION ON UPPER AND LOWER LIMITS RATHER THAN PRECISE PREDICTIONS. THE CONSTRAINTS TAKEN INTO CONSIDERATION ARE THE SIZE OF THE POPULATION AT RISK (AGES 20 TO 29) AND 'HISTORICAL GROSS EFFECTIVE POLICY LEVELS' (AS REFLECTED IN HIGH, MEDIUM, LOW RATES OF IMPRISONMENT). SUCH ANALYSIS SUGGESTS THAT THE POPULATION AT RISK WILL DECREASE ABOUT 20 PERCENT BETWEEN 1985 AND 1995 AND MAY INCREASE THEREAFTER. GROSS POLICY LEVELS MAY CHANGE BY 1985. IF POLICIES CHANGE AT THE SAME TIME THAT THE POPULATION AT RISK DECREASES, THE PRISON POPULATION COULD DECLINE SHARPLY BETWEEN 1985 AND 1995. HOWEVER, PRISON POPULATIONS WOULD CLIMB TO NEW HIGHS AFTER 1995. COMMENTS ON THE APPLICATION OF THE INTERMEDIATE-RANGE PROJECTION MODEL ARE OFFERED, TOGETHER WITH STRATEGY SUGGESTIONS FOR DEALING WITH FLUCTUATING PRISON POPULATIONS. GRAPHS AND A LIST OF REFERENCES ARE INCLUDED. (LKM)