Although one of the most important aims of physical education and public sport policy is to encourage life-long habitual physical activity, very little is known of the stability or tracking of physical activity. As a part of a larger research project called Cardiovascular Risks in Young Finns, the purpose of the study reported here was to investigate the stability (tracking) of leisure-time physical activity and sport participation at intervals of 3, 6, 9 and 12 years from age 9 to 30. The subjects, chosen by stratified random sample, represent five geographical areas of the country and eight gender-age cohorts (9, 12, 15 and 18 years old in 1980). Physical activity and sport participation were measured using a short questionnaire. Tracking was analysed by Spearman's rank order correlations and by simplex model. The tracking correlations within a 3-year interval of all individual variables were significant but rather low. The tracking correlations of the sum index of physical activity comprising five variables varied from 0.50 to 0.80 among boys and from 0.40 to 0.61 among girls. Simplex models that fitted the data very well showed higher stability coefficients than rank order correlations. With one exception the correlations of physical activity index within the 12-year interval were significant but low.