A CLASS OF SIMILARLY SITUATED OFFENDERS IS RANDOMLY DIVIDED INTO TWO GROUPS. ONE GROUP IS EXPOSED TO THE NEW TREATMENT (EXPERIMENTALS), WHILE THE OTHER IS TREATED AS IF NO EXPERIMENT WERE TAKING PLACE (CONTROLS). THE TWO GROUPS ARE THEN STUDIED AND COMPARED TO DETERMINE WHAT EFFECT, IF ANY, THE NEW TREATMENT HAS HAD AND WHETHER THAT EFFECT HAS PROVEN BENEFICIAL IN TERMS OF THE GOAL SET (E.G., REHABILITATION, DETERRENCE, PREVENTIVE DETENTION). THIS NOTE EXAMINES THE USE OF THIS RESEARCH TOOL IN THE CONTEXTS OF POST-CONVICTION TREATMENT OF OFFENDERS AND, TO A LESSER EXTENT, PRETRIAL DIVERSION. IN ADDITION, IT EXPLORES WHETHER SUCH EXPERIMENTS ARE PERMISSIBLE IN LIGHT OF EXISTING SOCIAL VALUES AND THE LIMITATIONS PLACED ON STATE ACTION BY THE CONSTITUTION. ANALYSIS OF BOTH CONSTITUTIONAL AND NONCONSTITUTIONAL ISSUES (DUE PROCESS, EQUAL PROTECTION, PROHIBITION OF CRUEL AND UNUSUAL PUNISHMENT) LEADS TO THE CONCLUSION THAT A PROGRAM MODEL BASED UPON THE THREE GUIDELINES OF ADMINISTRATIVE EXPERIMENTATION, LESSENED SEVERITY, AND NONEXCESSIVE DIFFERENCES IN TREATMENT MINIMIZES THESE PROBLEMS AND CREATES AN ACCEPTABLE RESEARCH TECHNIQUE. SUCH EXPERIMENTATION, HOWEVER, IS ONLY JUSTIFIABLE WHERE THERE IS A COMMITMENT TO PENAL REFORM. (AUTHOR ABSTRACT MODIFIED)...ELW