THE MINNEAPOLIS REHABILITATION CENTER (MRC) STUDIED MINNESOTA MULTIPHASIC PERSONALITY INVENTORY (MMPI) SCORES TO SEE IF THEY RELIABLY PREDICTED EMPLOYMENT OUTCOME AND TREATMENT EFFECTS. IT WAS DEEMED IMPORTANT FOR AGENCIES TO HAVE A METHOD OF DIAGNOSING PROBABLE RETURN-TO-WORK CLIENTS BEFORE SELECTING THOSE FOR REHABILITATION. A METHOD OF EVALUATING REHABILITATION SERVICES THROUGH MEASUREMENT OF CLIENT CHANGE WAS DEVELOPED. THE MMPI WAS ADMINISTERED TO THREE GROUPS (PILOT, CONTROL, AND CROSS-VALIDATION) UPON REFERRAL TO THE AGENCY AND FIVE WEEKS LATER. STATISTICAL ANALYSIS OF THE TEST RESULTS, MEDICAL AND DEMOGRAPHIC INFORMATION, AND THE WORK SITUATION OF THE CLIENTS PRODUCED NEGATIVE RESULTS. THE PROJECT DIRECTOR BELIEVES THAT THE STUDY FAILED TO PRODUCE RELIABLE RESULTS BECAUSE NONE OF THE PREDICTORS WERE DIRECTLY RELATED TO GOING BACK TO WORK. PROBLEMS OF GROUP SELECTION AND VARIABLE DEFINITION ARE DISCUSSED AS POSSIBLE REASONS FOR THE RESULTS. GUIDELINES FOR FUTURE RESEARCH ON SUCH PROBLEMS ARE SUGGESTED. THERE IS NOT YET AN ORGANIZED SYSTEM FOR MEASURING BEHAVIORS WHICH ARE IMPORTANT IN GETTING AND KEEPING A JOB. (NS)