IN 1975-1976 DURING THE 94TH CONGRESS, MORE THAN 30 SEPARATE BILLS CALLING FOR MANDATORY-MINIMUM PRISON SENTENCES WERE INTRODUCED. MOST OF THE BILLS LIMIT MANDATORY SENTENCING TO SPECIFIED CRIMES OR TO PARTICULAR CATEGORIES OF CRIMINALS. THE MOST COMMON TYPE OF CRIMINAL RECOMMENDED FOR MANDATORY SENTENCING IS THE REPEAT CRIMINAL, ON THE GROUNDS THAT THE SENTENCE SHOULD BE MADE SEVERE IN PROPORTION TO THE NUMBER OF PAST CONVICTIONS. SENTENCING REFORMS, HOWEVER, IMPLY GREATER CRIMINAL JUSTICE SYSTEMS COSTS, AND IT IS THEREFORE NECESSARY THAT SUCH REFORMS BE PRECEDED BY AN EVALUATION OF POTENTIAL BENEFITS IN CRIME REDUCTION AS COMPARED TO LIKELY COSTS ASSOCIATED WITH INCREASED PRISON POPULATIONS. ADVOCATES OF MANDATORY SENTENCING SCHEMES GENERALLY IGNORE THE PROJECTED EXTRA BURDEN ON THE CORRECTIONAL SYSTEM. POLICYMAKERS NEED TO KNOW WHAT TYPE OF OFFENDER SHOULD RECEIVE WHAT LENGTH OF SENTENCE TO PRODUCE THE LARGEST REDUCTION IN CRIME AND SECONDLY, WHAT IMPACT THESE MANDATORY PENALTIES WILL HAVE ON THE PRISON POPULATION. IT IS SUGGESTED THAT MANDATORY SENTENCING POLICIES CAN REDUCE CRIME AS A RESULT OF INCAPACITATION EFFECTS BUT THAT THE INCREASE IN THE PRISON POPULATION RESULTING FROM SUCH POLICIES MAY BE UNACCEPTABLY LARGE. TO REDUCE THE CRIME LEVEL BY HALF, EVERY PERSON CONVICTED OF A FELONY, REGARDLESS OF PRIOR CRIMINAL HISTORY, WOULD HAVE TO BE IMPRISONED FOR 5 YEARS. IF ONLY DEFENDANTS WHO HAVE A PRIOR ADULT CONVICTION ARE IMPRISONED, THE CRIME REDUCTION EFFECT IS ABOUT HALF THAT PRODUCED BY SENTENCING EVERY CONVICTED FELON TO PRISON. THE MOST EFFICIENT POLICY, IN THE SENSE OF PRODUCING THE HIGHEST CRIME REDUCTION AND THE LOWEST INCREASE IN THE PRISON POPULATION, APPEARS TO BE THAT OF SENTENCING ALL CONVICTED FELONS TO 1.2 YEARS OF PRISON. THIS POLICY REDUCES THE CRIME RATE BY 20 PERCENT WHILE RAISING THE PRISON POPULATION BY 85 PERCENT. JUDGES CAN BE SUCCESSFUL IN DISTINGUISHING AMONG DEFENDANTS WHO POSE MORE OR LESS SERIOUS RISKS TO THE COMMUNITY. MANDATORY-MINIMUM SENTENCING POLICIES THAT FOCUS ONLY ON DEFENDANTS WITH PRIOR RECORDS APPEAR TO BE LESS EFFECTIVE THAN POLICIES WHICH IGNORE PRIOR RECORDS. A LIST OF REFERENCES IS PROVIDED. (AUTHOR ABSTRACT MODIFIED)