Economic theory suggests that variations in countries age structure should affect the economy on an aggregate level. This paper investigates the relationship between age structure and GDP in 20 OECD countries using annual data from 1970 to 1999. Using new methodology, the relationship between the variables can be formulated in levels despite the presence of unit roots in the time series. Applying two panel cointegration tests proposed by Pedroni (1995, 1997a, 1999), support is found for a long run relationship between GDP and the number of people in five different age groups. Coefficient estimates from panel regressions support effects in line with the life cycle hypothesis and human capital theory; children and retirees are found to have a negative or relatively smaller positive effect on GDP than productive age groups.