Recent studies on African economic history have emphasized the structural impediments to African growth, such as adverse geographical conditions and extractive colonial institutions. The evidence is mainly drawn from cross-country regressions on late 20th century income levels, assuming persistent effects of historical causes over time. But to which extent has African poverty been a persistent phenomenon? Our study sheds light on this question by providing new evidence on long-term African growth-trajectories. We show that slave trade regressions are not robust for pre-1970s GDP per capita levels, or for pre-1973 and post-1995 growth rates. We calculate urban unskilled real wages of African workers in nine British African countries 1880-1965, adopting Allen’s (2009) subsistence basket methodology. We find that real wages were above subsistence level, rose significantly over time and were, in major parts of British Africa, considerably higher than real wages in Asian cities up to, at least the 1930s. We explain the intra-African variation in real wage levels by varying colonial institutions concerning land alienation, taxation and immigration.