The aim of the study was to investigate the relationship between the Flynn effect and the effects of age differences between siblings on the intelligence difference between them. In Norway, the secular trends in intelligence-test score means vary both in magnitude and direction. We identified three periods: one period where the mean intelligence increased across birth cohorts (1950-1956), one period where the means decreased (1976-1983), and one period with no appreciable Flynn effect (1960-1965). In a data base comprising birth year and intelligence data on more than 900,000 males meeting at mandatory conscription between their 18th and 21st birthday, we identified more than 69,000 brother pairs where both brothers had been born in exactly one of the periods mentioned above. In this study group the relationship between age differences between brothers and the intelligence difference could be studied. The results showed that in the period with increasing intelligence means across cohorts, the intelligence difference between brothers decreased with increasing age differences. In the period with decreasing means, the difference between the later-born and the earlier-born brother increased across age differences. No systematic effects of age difference on mean intelligence differences were found in the period without a Flynn effect. Regression analyses showed that the Flynn effect can be quite well predicted from the effects of the age differences between brothers on their intelligence-test scores. We conclude that the factors causing the Flynn effect also work within sibships. Hypotheses positing that the Flynn effect is solely caused by between-families factors (e.g. the heterosis hypothesis) are weakened. The present results also entail that the birth order effect observed in Norway is in part conditional on the Flynn effect. (Contains 3 figures and 1 table.)