TO TEST THE VALIDITY OF RETROSPECTIVE PRETESTING IN THE EVALUATION OF EDUCATIONAL PROGRAMS, TWO TESTS WERE ADMINISTERED TO PARTICIPANTS IN A SUMMER COURSE FOR AMERICAN STUDENTS WHO, HAVING ATTENDED FOREIGN MEDICAL SCHOOLS, WERE SEEKING TO TRANSFER TO SCHOOLS IN THE UNITED STATES. BEFORE AND AFTER THE COURSE, THE STUDENTS ALSO COMPLETED A SECOND TEST, WHICH INCLUDED TWO SECTIONS: (1) 46 QUESTIONS FROM THE PREVIOUSLY ADMINISTERED NATIONAL TEST; AND (2) 105 QUESTIONS BASED ON MATERIAL ACTUALLY COVERED IN THE COURSE. IN BOTH SECTIONS, STUDENTS ANSWERED EACH QUESTION TWICE: ONCE WITH THE RESPONSE THEY THOUGHT CORRECT AND ONCE AS THEY THOUGHT THEY WOULD HAVE RESPONDED BEFORE TAKING THE COURSE. ANALYSIS OF TEST RESULTS FOR 34 STUDENTS INDICATES THAT IT IS POSSIBLE TO INFER LEARNING FROM STUDENTS' RECOLLECTION OF PREVIOUS KNOWLEDGE. THE VALIDITY OF RETROSPECTIVE PRETESTING SHOULD BE TESTED FURTHER IN A VARIETY OF LEARNING SITUATIONS. ONCE VALIDITY IS ESTABLISHED, RETROSPECTIVE PRETESTING MAY SOLVE THE PROBLEM OF ATTRIBUTING CHANGE TO EDUCATIONAL PROGRAMS THAT ARE OPEN, LOOSELY PLANNED, OR HIGHLY RESPONSIBLE TO STUDENT NEEDS OR DEMANDS. RETROSPECTIVE PRETESTING MAY ALSO SERVE AS A SUBSTITUTE FOR MORE COSTLY TIME-SEQUENTIAL SAMPLING PROCEDURES. SUPPORTING DATA AND A LIST OF REFERENCES ARE INCLUDED. (LKM)