Based on a sample of 1,084 European regions (EU 15) over the period of 1995-2004, we estimate the determinants of regional growth of GDP per capita, allowing for both spatial lag and spatial error dependence. We find that robust LM tests cannot reject the null hypothesis of no spatial dependence when country dummy variables are included in the growth equation. OLS and robust regression methods show that population density and industry share are significantly and positively related to economic growth. Regions that received EU structural funds have a significantly higher growth of GDP per capita, but the effect is only marginally significant. Blinder-Oaxaca decompositions reveal that the growth differential between Objective 1 regions and the remaining regions is solely due to the difference in the characteristics and not to differences in the coefficients. Finally, we find that the added value gained in Objective 1 regions is much lower than the resources that have been allocated to them.