THE INFLUENCE OF FOUR VARIABLES (SEX OF VICTIM, CONVERSATION, SIZE OF BYSTANDER GROUP, AND LOCATION) ON THE BEHAVIOR OF NAIVE BYSTANDERS TO A CRIMINAL SITUATION WERE EXAMINED. SEVEN MALES AND SEVEN FEMALES PLAYED THE ROLE OF VICTIMS. ON HALF OF THE EXPERIMENTAL TRIALS, THE VICTIM-TO-BE INITIATED A BRIEF CONVERSATION WITH THE NAIVE SUBJECT OR SUBJECTS, USUALLY BY INQUIRING ABOUT THE CORRECT TIME. ON TRIALS WITHOUT CONVERSATION, NO VERBAL INTERACTION WAS PERMITTED PRIOR TO THE STAGED INCIDENT. SUBJECTS WERE STUDIED INDIVIDUALLY OR IN NATURAL GROUPS OF TWO OR MORE. THE STUDY WAS CONDUCTED IN THREE LOCATIONS AT A LARGE MIDWESTERN UNIVERSITY. ANALYSIS OF VARIANCE SHOWED THAT ALL VARIABLES WERE SIGNIFICANT. SUBJECTS WERE MORE WILLING TO INTERVENE IN AN APPARENT BOOK THEFT IF THE VICTIM WAS A FEMALE AND IF THE VICTIM HAD INITIATED A CONVERSATION PRIOR TO THE THEFT. THE MAIN EFFECT OF LOCATION INDICATED THAT MORE INTERVENTION OCCURRED IN THE STUDENT LOUNGE THAN IN THE LIBRARY OR GRILL WHERE MEAN INTERVENTION SCORES WERE ABOUT EQUAL. TWO SIGNIFICANT INTERACTIONS, SEX OF VICTIM BY LOCATION AND INTERACTION OF THESE TWO FACTORS WITH CONVERSATION, WERE ALSO OBTAINED. ALTHOUGH THE EXPERIMENT WAS NOT DESIGNED TO SERVE AS A CRITICAL TEST OF COST-DEPENDENCY AND DIFFUSION OF RESPONSIBILITY ORIENTATIONS, THE RESULTS CAN BE USED TO EVALUATE THE UTILITY OF THESE EXPLANATIONS IN PREDICTIVE MODELS OF BYSTANDER INTERVENTION. SUPPORTING DATA AND REFERENCES ARE INCLUDED. (DEP)