While the concerns about adolescent social media use and its effect on life satisfaction are rising, scientific evidence in the area is still surprisingly sparse: with most studies being purely cross-sectional and unable to dissociate between-person associations and individual longitudinal relations. While comparisons of different people simultaneously are inherently different from examinations of the same person longitudinally, many previous studies erroneously interpret the former as the latter. Disentangling these estimates, this study uses high-quality, large-scale and nationally representative panel data to examine the relationships between social media use and life satisfaction. Implementing Specification Curve Analysis and Random Intercept Cross-lagged Panel Models, it shows that the median cross-sectional association between social media use and life satisfaction is negative. The individual longitudinal relationships are also negative, but so small they merit little consideration. It is however in evidence that gender is a substantial and important determinant of the link between social media use and well-being. For adolescent boys there is little evidence for cross-sectional relations, or longitudinal effects. For adolescent girls, however, increases in life satisfaction predicts lower social media use one year later and increases in social media use also predicts lower life satisfaction one year later. Furthermore, those girls who use more social media than other girls show lower life satisfaction. Social media is therefore not inherently bad for the totality of the adolescent population, but is instead contingent on gender and data analysis techniques. Research and policy discussion needs to therefore move focus to the motives and modes young women adopt when using social media.