OBJECTIVE:To investigate relationships between frequency of physical activity or television viewing and body mass index (BMI) cross-sectionally at six ages from childhood to adulthood, to better understand longitudinal relationships. To investigate how the relationships vary with age and gender and whether any relationships are due to confounding factors. METHODS: The 1958 British birth cohort includes all births (approximately 17 000) in one week in March 1958. BMI and physical activity frequency were recorded at 11, 16, 23, 33 and 42 y and television viewing frequency at 11, 16 and 23 y. A total of 11 109 subjects provided BMI and activity data at 42 y. Relationships between BMI and (in)activity were investigated using linear regression. RESULTS: At ages 11, 33 and 42 y in both sexes and at 23 y in female subjects, those who were more active had lower BMIs, and the relationships strengthened with age. At 42 y, the most active had a lower mean BMI than the least active, by 0.83 kg/m 2 in men, and 1.03 kg/m 2 in women. BMI and activity were unrelated at 16 y in female subjects, and 23 y in male subjects. At 16 y in males, the most active males had a mean BMI 0.25 kg/m 2 higher than the least active. At 11 y in female subjects and 23 y in both sexes, those who watched television most frequently had higher BMIs. BMI and television viewing were unrelated at 11 y in males and at 16 y in both sexes. Relationships between BMI and activity or television viewing were largely unexplained by potential confounding factors. CONCLUSIONS: The relationship between BMI and physical activity changes with age. In early adolescence and in adulthood, a higher activity level, or lower frequency of television viewing was associated with a lower BMI. In later adolescence (16 y), television viewing and activity were unrelated to BMI, except for an unexpected BMI-activity relationship in males. We suspect this relationship in males is primarily due to selection effects, whereby physically bigger boys, with a larger BMI, are more likely to take part in exercise activity, and possibly also to BMI being a less accurate predictor of fatness in adolescent boys.