The proposition that institutions explain why some countries are rich has become close to orthodoxy, and its evidential base is narrower than the confidence with which it is asserted. The influential studies rely on instruments — colonial settler mortality, the identity of the colonising power — designed to isolate institutional quality from everything else, and the plausibility of each instrument has been contested on grounds that are not easily dismissed, since the same historical facts plausibly influenced growth through channels other than institutions. What survives the criticism is a correlation of considerable strength and a causal interpretation resting on assumptions rather than on any experiment. Believing that institutions matter is entirely reasonable. Believing that we have measured how much they matter is a different proposition, and one the literature supports less firmly than its summaries suggest.