RESEARCHERS CONDUCTING CONTROL GROUP EXPERIMENTS IN FIELD SETTINGS WHERE THE TREATMENT INVOLVES CLIENT SERVICES OFTEN ENCOUNTEER OPPOSITION BY CLIENTS AND PRACTITIONERS WHEN CLIENTS MUST BE ASSIGNED RANDOMLY TO EXPERIMENTAL AND CONTROL GROUPS, DUE TO THE EXCLUSION OF CONTROL GROUP MEMBERS FROM TREATMENT. THE USE OF COMPARISON GROUPS INSTEAD OF CONTROL GROUPS AND THE USE OF DELAYED TREATMENT FOR CONTROLS ARE TWO ALTERNATIVE SELECTION PROCEDURES USED. THE ADVANTAGES AND DISADVANTAGES OF THESE ALTERNATIVES ARE DISCUSSED. A THIRD ALTERNATIVE, WHICH IS THE PRINCIPAL SUBJECT OF THIS PAPER, WAS USED IN THE EVALUATION OF A PRETRIAL DIVERSION PROGRAM THAT CHANNELED DEFENDANTS INTO EMPLOYMENT SERVICES, EDUCATIONAL COUNSELING, AND OTHER REFERRAL SERVICES. BECAUSE OF LIMITED FUNDS, THE PROGRAM WAS NECESSARILY RESTRICTED TO A CERTAIN NUMBER OF CLIENTS. THE SELECTION PROCEDURE PROVIDED FOR THE SCREENING OUT ON THE BASIS OF PROGRAM CRITERIA MORE DEFENDANTS THAN COULD BE DIVERTED TO THE PROGRAM, THUS PROVIDING AN OVERFLOW OF ELIGIBLE SUBJECTS TO SERVE AS A CONTROL GROUP. A QUOTA SYSTEM WAS DEVELOPED IN ORDER TO SELECT THOSE CASES TO BE DIVERTED DURING A GIVEN TIME PERIOD (E.G., A MONTH, WEEK, OR DAY). OVER AN EXTENDED PERIOD, THESE QUOTAS WOULD EQUAL THE TOTAL NUMBER OF CASES THE PROJECT HAD FUNDS TO SERVE. THE REMAINING CASES SCREENED AND APPROVED DURING ANY TIME PERIOD WOULD CONSTITUTE AN OVERFLOW OF ELIGIBLE DEFENDANTS TO BE PROCESSED NORMALLY BY THE COURT (CONTROL GROUP). THE PROCEDURE FOR DETERMINING THE QUOTA PERIODS USED IN THE RESEARCH ASSIGNMENT IS DESCRIBED. THE EXPERIMENTAL AND CONTROL GROUPS THAT RESULTED WERE COMPARED ON SOCIOECONOMIC AND CRIMINAL BACKGROUND AND CHARACTERISTICS OF THEIR COURT CASES AND FOUND TO BE SIMILAR. SOME GUIDELINES ARE SUGGESTED FOR THE USE OF THIS SELECTION TECHNIQUE, AND ITS COMPARATIVE ADVANTAGES IN RELATION TO OTHER ALTERNATIVES TO RANDOM SELECTION ARE DISCUSSED.