A SURVEY OF STUDIES PUBLISHED IN SOCIAL WORK RESEARCH AND ABSTRACTS FROM 1975 THROUGH 1978 IDENTIFIED 97 EVALUATION STUDIES. THE LARGEST GROUP (25 PERCENT) WAS CONDUCTED IN CHILD WELFARE SETTINGS. ABOUT HALF WERE EVALUATIONS OF DEMONSTRATION PROJECTS OR EXPERIMENTS; THE OTHER HALF, OF STANDARD PROGRAMMING. FOUR-FIFTHS OF THE STUDIES LACKED COMPARISON GROUPS; ABOUT 10 PERCENT USED NONEQUIVALENT COMPARISON GROUPS; AND 10 PERCENT USED EQUIVALENT GROUP COMPARISONS WITH RANDOMLY ASSIGNED SUBJECTS. SURVEYS INDICATE THAT EVALUATION RESEARCH IS A GROWTH INDUSTRY IN SOCIAL WORK. TWENTY-TWO STUDIES DESIGNED TO TEST SOCIAL WORK EFFECTIVENESS HAVE BEEN CONDUCTED BETWEEN 1956 AND 1973. ALTHOUGH MOST PRODUCED NEGATIVE OR EQUIVOCAL RESULTS, CRITICS HAVE CHARGED THAT THE EVALUATION METHODS USED WERE FLAWED. IT IS AGREED THAT SUCH GROSS EVALUATIONS OF AMORPHOUS SERVICE METHODS, SUCH AS CASEWORK, SHOULD BE AVOIDED. A SEARCH FOR BETTER EVALUATION METHODS IS UNDERWAY. ABOUT A DOZEN RECENT GROUP FIELD EXPERIMENTS SHOW THAT INTERVENTION VARIABLES ARE BEING MORE NARROWLY AND BETTER DEFINED, SERVICE GOALS AND OUTCOMES MORE CIRCUMSCRIBED, AND RESEARCH DESIGN MORE COMPLEX. THESE STUDIES HAVE TENDED TO PRODUCE POSITIVE FINDINGS. SINGLE SUBJECT RESEARCH DESIGNS ARE ALSO BEING ADVOCATED, BUT VIABLE PROGRAM MODELS USING THIS METHOD HAVE YET TO APPEAR IN SOCIAL WORK. ANOTHER APPROACH IS DEVELOPMENTAL RESEARCH. IT HAS BOTH ADVANTAGES AND DISADVANTAGES, BUT CANNOT REPLACE MORE USUAL FORMS OF PROGRAM EVALUATION. AGENCY ACCOUNTABILITY SYSTEMS USING ROUNTINELY COLLECTED DATA ARE PROBABLY MOST USEFUL IN EVALUATING PROGRAMS WHEN RELEVANT INDICATORS CONSIST OF READILY MEASURABLE AND VERIFIABLE PHENOMENA. IT IS CONCLUDED THAT EVALUATION RESEARCH IN SOCIAL WORK IS BECOMING MORE VARIED AND IS SHOWING SIGNS OF GREATER SOPHISTICATION THAN IN THE PAST. PROGRESS MAY SLOW IN THE NEAR FUTURE, BUT MAY BE OF MORE SIGNIFICANCE WHEN IT DOES OCCUR. A REFERENCE LIST IS INCLUDED. (CFW)