This paper shows the evaluation of the effectiveness of a selective prevention on drug use in adolescents, based on the Theory of Planned Behavior by Azjen. To develop this program, different theoretical models were evaluated considering their suitability to the institutional context in which it will be implemented. The criteria used to select a model that could be useful as a basis to design and evaluate a preventive intervention for drug use included: a) incorporation of protective and risk factors that could be malleable; b) consistency, parsimony and theoretical economy; c) a demonstrated explanatory and predictive capacity; d) empirical effectiveness; e) practical applicability. One of the models that complies with these requirements is the Model of Planned Behavior by Azjen, which proposes that the most immediate causes for drug use are the intentions of the subject to consume them or not. These intentions are determinated by the «attitudes» toward drug use, by the normative beliefs or «subjective norm» and by the Perceived Behavior Control. Attitudes toward the behavior are a person's overall evaluation of the behavior (i.e drug use) and the corresponding positive or negative judgments about it. The normative beliefs or subjective norms are a person's own estimate of social pressure to use or not to use drugs. Subjective norms have two interacting components: beliefs about how other people, who may be significant to the person, would like them to behave (normative beliefs), and the positive or negative judgments about each belief (outcome evaluations). Perceived behavioral control is the magnitude to which a person feels able to carry out the behavior. It includes two aspects: how much control a person has over the behavior; and how confident a person feels about being able to perform or not the behavior. It is determined by control beliefs about the power of both situational and internal factors to inhibit or facilitate the performance of the behavior. Perceived behavioral control, the last element introduced in the theory, along with the attitude to conduct, and the subjective norm, contribute, all of them, to predict the behavior intention. Generally speaking, there is important empirical evidence concerning the predictive and explicative capacity of this model in different populations. Once the theoretical model was chosen, indicators related to constructs of the Theory of Planned Behavior were elaborated adapting these constructs to the features of middle education Mexican population by means of a series of interviews with focal groups. Based on these indicators, a scale was constructed to prove the explanatory capacity of the theoretical model in Mexican population. The reliability and validity of this self-applied questionnaire was proved. Once this was done, based on these antecedents, the model was adapted in a practical sense. This means that the content, objectives and techniques for a preventive intervention were developed. This preventive intervention was piloted and then evaluated through the aforementioned scale. The program included attitudinal and normative components, conformed by informative strategies and persuasive messages directed to an attitudinal change, as well as the development of strategies of behavioral control focused on social abilities that included components derived from social learning theory, and a series of cognitive behavioral techniques accorded to the Planned Behavioral Theory. It was expected a reduction in the drugs use intention by means of a change of drug use favorable beliefs as well as reinforcing unfavorable beliefs of drug use. It was also tried to change the perception about the magnitude of drug users in the subject's immediate surroundings and on its perception about social acceptance or tolerance for drugs. Also it was tried to increase the perception of self control, in order to the develop behavioral control skills to refuse the offer of drugs. The first sessions (devoted to work on subjective norm) were directed to create an interest in the participants regarding to their relation with favorable situations to drug consumption, and to modify their perception about the real magnitude of drug use and about approval and social tolerance to them, both topics frequently overestimated for some adolescents. These sessions were also devoted to identifying beliefs associated with drug use and the value that participants give to these beliefs, in order to question those that are established upon false premises. The next sessions (devoted to perceived behavioral control) had as an objective to learn appropriated behavioral responses to confront risk situations, in order to increase the perception of self-control in those circumstances. This was achieved by means of modeling, essay and reinforcing techniques. This model was designed to be applied in groups no greater than 15 persons, twice a week in five sessions of 50 minutes each. The program was applied by professional personnel such as psychologists or social workers with previous experience in drugs use prevention. The program was implemented in three groups (one for each grade) of five high schools in Mexico City and was applied to three groups of school, randomly chosen; other three groups, in the same grades, were used as control groups. The groups were randomly chosen. Other three groups, in the same grades, were used as comparison groups. Then, the final sample was composed by 250 participants (96 from intervened groups and 154 from comparison groups) from 12 to 15 years old, and with slightly more men than women, but equivalent for cases and control. The results showed that the intervened group had a statistical significant reduction in the intention of drugs use. On the contrary, the comparative group showed an increasing intention to use them. With respect to the set of variables that predicted the behavioral intention, the perceived behavioral control stands out as the variable that had the greatest change. There were no significant differences in pre-post measures in the control group. Differences between pre-post measurements were also contrasted between both groups using a Student's t test that showed significant differences in perceived behavioral control and behavioral intention variables. Finally, in order to get a greater control on the test scores, an ANCOVA analysis was applied using the post-test scores as dependent variable and the pre-test scores as covariables. This analysis was useful to corroborate the significance of the drug use intention in the future, mainly in situations that require an amount of self-control, while attitude and normative variables did not show significative changes. Outcomes confirmed the pertinence to give greater resources (time, quantity, and complexity of activities, didactic materials, etc.) to the sessions devoted to developing behavioral control skills. These results can be considered satisfactory, because the main objective of the program was to reduce the intention to use drugs; however, it is also necessary to strengthen the attitudinal and normative components. It is necessary to consider that the variable «subjective norm» obtained the lowest levels in confidence, possibly because of the use of self-reported answers in a variable so sensitive to the influence of social elements. For this reason, it is necessary to develop questions that diminish this element that, according to Doll and Azjen, can explain, even a 5% of variance in «behavioral intention». Finally, it must be considered that this evaluation shows the effectiveness of the program to achieve the expected changes in behavioral intention variable only in controlled situations. It is necessary to investigate if it is possible to obtain similar results in more usual conditions and with other populations. This means that as well as to grant the internal validity of the evaluation it is also necessary to establish its external validity. [ABSTRACT FROM AUTHOR]