This paper combines a new historical data set regarding capital formation in infrastructure in the Netherlands in the nineteenth century with data-oriented econometric techniques aimed at testing the causal relationship between these infrastructural investments and economic growth. The resulting vector autoregression (VAR) model is analysed further with impulse response analysis. The results show that the time series characteristics of both capital formation and GDP are trend stationary, which is a fundamental difference with their twentieth century counterparts. The paper finds strong evidence of both a long term and medium term or short term impact. In the short run positive expenditure effects are partly offset bu negative transitional dynamics. To fine tune the analysis we have exploited the possibility to disaggregate the data set in basic and complementary infrastructure investment. Whereas the effect on output is significantly positive for basic infrastructure investment, it is absent for complementary infrastructure investment.