IT IS CAUTIONED THAT THE ADMINISTRATIVE PROCEDURES USED IN DECISIONMAKING ABOUT EVALUATION IN CONTROLLED PROGRAMS CANNOT BE APPLIED IN UNREVISED FORM TO OCCASIONS FOR EVALUATION RESEARCH ASSOCIATED WITH 'NATURAL EXPERIMENTS.' 'NATURAL EXPERIMENTS' ARE DEFINED AS UNPLANNED OR UNANTICIPATED OCCURRENCES WHICH SUPPLY DATA AND HAVE IMPACTS OF CONCERN TO SOCIAL SCIENTISTS. INCIDENTS OF THIS TYPE IN THE FIELD OF ALCOHOL STUDIES GENERALLY CONCERN ALCOHOL ABUSE CONTROL MEASURES. THE IMPACT OF LAWS REGARDING THE LEGAL LIMIT OF THE BLOOD-ALCOHOL CONTENT WHILE DRIVING, THE MINIMUM AGE FOR ALCOHOL CONSUMPTION, AND THE PLACING OF OF ALCO-METERS IN TAVERNS (I.E., FOR CUSTOMERS TO MEASURE THEIR BLOOD-ALCOHOL CONTENT SO AS TO INFLUENCE THEIR DECISION ABOUT WHETHER TO DRIVE FROM THE TAVERN) ARE CITED AS 'NATURAL EXPERIMENTS' BECAUSE THEY INVOLVE ACTIONS BY LAWMAKERS THAT CANNOT REASONABLY BE ANTICIPATED OR PLANNED FOR. THESE OCCASIONS MAY REQUIRE THE DECISION TO MAKE A STUDY ON SHORT NOTICE, ALTHOUGH THAT IS INCONSISTENT WITH THE INCREASING FORMALIZATION OF THE STRUCTURE OF EVALUATION RESEARCH. RESOLUTION OF THIS PROBLEM LIES IN ADJUSTING THE STRUCTURE OF EVALUATION RESEARCH TO MEET THE NEED OF THE RESEARCHER TO CAPTURE A RARE OPPORTUNITY AND THE NEED OF THE ADMINISTRATOR TO PLAN AHEAD. IT IS SUGGESTED THAT IMPORTANT THEORETICAL QUESTIONS CAN BE ANSWERED BEFORE THE OCCURRENCE OF A 'NATURAL EXPERIMENT.' BROAD CLASSES OF SUCH EVENTS CAN BE IDENTIFIED, AND DIFFERENT RESEARCH STRATEGIES AND COST ESTIMATES CAN BE ASSOCIATED WITH EACH CLASSIFICATION. BUDGETING AND OTHER PLANNING ACTIVITIES WOULD THEN ALLOW FOR A REALIGNMENT OF PRIORITIES IN A RELATIVELY SHORT TIME ON THE OCCASION OF A 'NATURAL EXPERIMENT.' REFERENCES ARE PROVIDED. (RCB)