Differences in institutions are thought to be a key part of the explanation for large cross-country differences in per capita income. Unfortunately, recent empirical research suffers from severe endogeneity problems. In this paper we provide both a theory of why different countries have different institutions and a way of measuring this, which does not suffer from endogeneity. We focus on economies that were colonized by European countries between the sixteenth and nineteenth centuries. Institutions in these countries were shaped by their colonization experience and we can use the differences in these experiences as an exogenous source of variation in institutions. We focus on differences in state institutions that depended crucially on settlement patterns. We use data on the mortality of settlers during the eighteenth and nineteenth centuries as an instrument for current institutional differences and show that institutions are a major determinant of different income levels. In our basic empirical model, once we instrument for institutions, two-thirds of the variation in current per capita income is explained by them.