'ACTION RESEARCH' IS DEFINED AS WORKING TOGETHER WITH THE PEOPLE IMMEDIATELY INVOLVED, TOWARDS A PROCESS OF CHANGE AND TO ACCOMPANY THAT PROCESS WITH RESEARCH. THE ACTION RESEARCH CONSIDERED IN THIS PAPER WAS ATTEMPTED IN A PSYCHIATRIC HOSPITAL FOR 72 ADULTS, BOTH MALE AND FEMALE, COMMITTED FOR TREATMENT BY A CRIMINAL COURT 'MEASURE' (CONSTRASTED WITH PUNISHMENT IN DUTCH CRIMINAL LAW). BASED ON HIS EXPERIENCES OBSERVING ACTION RESEARCH IN AMERICA, THE DIRECTOR OF THE HOSPITAL SOUGHT TO PUT ACTION RESEARCH INTO EFFECT IN THE HOSPITAL IN 1974. GOAL ATTAINMENT SCALING (GAS) WAS INTRODUCED, STARTING WITH NEW PATIENTS. THIS METHOD, DESIGNED TO EVALUATE INDIVIDUAL TREATMENT PROGRAMS CONSISTS OF A DESCRIPTION OF THE PATIENT'S PROBLEM, A DELINEATION OF THE TREATMENT GOAL, AND A DESCRIPTION OF THE TREATMENT METHOD. PROBLEMS WITH THE IMPLEMENTATION OF THE GAS WERE FOUND TO BE AS FOLLOWS: DIFFICULTY OF ARRIVING AT AN UNDERSTANDING OF THE PROBLEM AND TREATMENT GOALS WITH PATIENTS, COOPERATION WITH TREATMENT STAFF AND RESEARCH WORKERS WAS DIFFICULT TO OBTAIN, THE SHIFT SYSTEM PREVENTED EFFECTIVE CONTINUOUS STAFF COOPERATION, AND THE TREATMENT STAFF OFTEN FAILED TO ADHERE TO DETERMINED GOALS AND METHODS. METHODOLOGICAL PROBLEMS WITH THE GAS INCLUDED: CONFUSION AS TO WHETHER THE PATIENT OR THE TREATMENT STAFF IS THE PARTNER OF THE RESEARCH WORKER, PROBLEMS ASSOCIATED WITH COLLECTING AND ORDERING DATA, CONCEPTUALIZATION BY THE RESEARCH WORKER OF WHAT WAS HAPPENING IN THE TREATMENT PROCESS WAS FREQUENTLY CONSIDERED BY TREATMENT PERSONNEL TO BE IRRELEVANT TO ACTUAL TREATMENT SITUATIONS, AND EVALUATION FREQUENTLY REVEALS THAT GOALS WERE NOT FORMULATED PRECISELY ENOUGH TO PERMIT MEANINGFUL EVALUATION. IT IS GENERALLY CONCLUDED THAT THE SUCCESS OF ANY ACTION RESEARCH PROJECT DEPENDS UPON A COLLABORATION BETWEEN PATIENTS, TREATMENT WORKERS, AND RESEARCHERS WHICH AIMS AT IMPROVING THE LOT OF THE PATIENT. REFERENCES ARE PROVIDED. (RCB)