THE EXPERIMENT HYPOTHESIZED THAT FOR BOTH ALCOHOL AND DRUGS THE PROBABILITY OF ABSTINENCE DECREASES AND THE FREQUENCY OF USE INCREASES WHEN ADOLESCENTS ASSOCIATE MORE WITH USERS THAN WITH ABSTINENT PEERS AND ADULTS, WHEN USE IF DIFFERENTIALLY REINFORCED (MORE REWARDS, FEWER PUNISHERS) OVER ABSTINENCE, AND WHEN DRUG USE IS DEFINED IN MORE POSITIVE OR NEUTRAL TERMS THAN NEGATIVE. THE PROBABILITY OF ABUSE WAS ALSO THEORIZED TO FOLLOW THE SAME PATTERN. DATA WERE COLLECTED BY ADMINISTERING A SELF-REPORT QUESTIONNAIRE TO 3,065 MALE AND FEMALE ADOLESCENTS ATTENDING GRADES 7 THROUGH 12 IN 7 COMMUNITIES IN 3 MIDWESTERN STATES. A TWO-STAGE SAMPLE DESIGN WAS FOLLOWED. THE SCHOOLS SELECTED FROM EACH PARTICIPATING SCHOOL DISTRICT WERE REPRESENTATIVE IN SIZE AND LOCATION IN THE DISTRICT. THE QUESTIONNAIRE (PRETESTED IN A DISTRICT NOT INCLUDED IN THE FINAL SAMPLE) WAS ADMINISTERED TO ALL STUDENTS IN ATTENDANCE ON THE DAY OF THE SURVEY WHO HAD OBTAINED WRITTEN PARENTAL PERMISSION. ABSTINENCE USE OF ALCOHOL AND MARIJUANA WAS MEASURED BY A SIX-POINT FREQUENCY-OF-USE SCALE RANGING FROM NEARLY DAILY USE TO NEVER. COMBINATIONS OF RESPONSES TO THE PREVIOUS SCALE PRODUCED A FOUR-POINT ABUSE SCALE RANGING FROM HEAVY ABUSE TO NO ABUSE. THE MAIN CONCEPTS MEASURED WERE IMITATION, DIFFERENTIAL ASSOCIATION, DEFINITIONS, AND DIFFERENTIAL REINFORCEMENT. WHEN THE FIVE SETS OF INDEPENDENT VARIABLES WERE INCORPORATED INTO THE FULL REGRESSION EQUATION, THE MODEL EXPLAINED 55 PERCENT OF THE VARIANCE IN DRINKING BEHAVIOR AND 68 PERCENT OF THE VARIANCE IN MARIJUANA USE. THE POWER OF THE FULL SOCIAL LEARNING MODEL, THEREFORE, WAS DEMONSTRATED. A LIST OF 33 REFERENCES IS PROVIDED.