LEGAL THREATS DO NOT OPERATE IN A VACUUM. THERE IS EVIDENCE FROM STUDIES ON MARIHUANA USE AND OTHER BEHAVIORS TO INDICATE THAT INFORMAL CONTROLS (E.G., PEER PRESSURE) AND SOCIAL-STRUCTURAL VARIABLES ARE MORE IMPORTANT DETERMINANTS OF DEVIANCE OR CONFORMITY THAN ARE FORMAL SANCTIONS. LEGAL THREATS ENCOMPASS A SMALLER REALM OF HUMAN BEHAVIOR THAN DO OTHER TYPES OF INFLUENCES. IN ADDITION, THERE APPEAR TO BE SOME GROUPS (E.G., YOUNG PEOPLE) WHO ARE RELATIVELY UNLIKELY TO RESPOND TO LEGAL THREATS. EVEN THOUGH ANY DETERRENT EFFECT OF LEGAL THREATS IS LIKELY TO BE A SMALL ONE, POLICYMAKERS MIGHT STILL BE INTERESTED IN MAXIMIZING WHATEVER EFFECT IS POSSIBLE. THE MOST OBVIOUS APPROACH TO MAXIMIZING DETERRENCE IS TO MANIPULATE THE THREE CLASSIC CONDITIONS OF DETERRENCE--SEVERITY, CERTAINTY, AND SWIFTNESS OF PUNISHMENT. SEVERITY--THE EASIEST CONDITION TO MANIPULATE--APPEARS TO BE THE LEAST LIKELY TO YIELD DETERRENCE. PRACTICAL AND OPERATIONAL BARRIERS IN CRIMINAL JUSTICE AGENCIES HAMPER MANIPULATION OF CERTANITY AND SWIFTNESS. THERE IS ANOTHER, APPARENTLY MORE PROMISING FOCUS FOR MANIPULATION: PERCEIVED CERTAINTY OF PUNISHMENT. CRIME CONTROL PROGRAMS THAT HAVE REPORTED SIGNIFICANT, MEASURABLE DETERRENT EFFECTS HAVE BEEN THOSE THAT CONCENTRATED ON, OR AT LEAST PRODUCED AS A BYPRODUCT, CHANGES IN THE PERCEIVED RISK TO POTENTIAL VIOLATORS. MANY FACTORS LIMIT THE DETERRENT EFFECT ANY CRIME CONTROL PROGRAM IS LIKELY TO HAVE. BUT THE LIMITATIONS OF DETERRENCE ARE NOT MERELY PRAGMATIC. THE DOCTRINE ITSELF HAS ETHICAL AND MORAL LIMITS. IT ASKS WHETHER CRIME DECREASES AS CERTAINTY OF PUNISHMENT INCREASES, BUT NOT WHETHER CERTAINTY OF PUNISHMENT SHOULD BE INCREASED IN ORDER TO DECREASE CRIME. DETERRENCE DOCTRINE CANNOT INDICATE WHETHER A GIVEN POLICY IS POLITICALLY FEASIBLE OR ETHICALLY DEFENSIBLE. DETERRENCE RESEARCH MAY ACTUALLY PROVE WORTHLESS WITH RESPECT TO CRIME CONTROL POLICYMAKING. A LIST OF REFERENCES IS INCLUDED. (LKM)