Résumé : In order to define the Rosenman and Friedman type A pattern three technics where used: the Bortner scale, the Jenkins Acitivity Survey (J.A.S.) and the interview. These are part of the base-line data recorded during the screening of a belgian male population aged 40 to 59 at entry of a controlled trial for the multifactorial prevention of cardiovascular diseases. The J.A.S. and the Bortner scale are evaluated in relation to the interview. Extremes of type A and B are less prevalent in our study compared to the Western Collaborative Group Study. Both the J.A.S. and the Bortner scale are very satisfactory in predicting type A and B pattern. The Bortner scale predicts the pattern in 78% of the subjects, by the method of weighted scores. The validation of this classification method, on a second sample, gives an accurate classification in 75% of the subjects. The J.A.S. predicts, the pattern in relation to the interview in 78% of the subjects; again the validation of the classification method, on a second sample, gives a correct classification in 70% of the subjects.