This paper discusses the trend in child mortality during the British colonial period in Tanzania from an inter-disciplinary perspective, suggesting how micro- and macro-demographic approaches could be combined with historical analysis to study pre-independence demographic trends in sub-Saharan Africa. METHODS: Estimates were made of the national trend in child mortality by applying Brass’s indirect techniques to a model based on the reconstruction of fertility from all census and survey data from 1948 to 1999 and by fitting model life tables using the reported total proportions of children dying in the 1948 and 1957 sample censuses. RESULTS: Contrary to what was reported in the censuses at the time, and contrary to the conventional expectations of demographic transition, fertility appears to have been rising between the mid-1940s until the 1970s. The model of child mortality suggests rates were steady during the interwar period and did not begin to decline until the 1950s. DISCUSSION: The finding of no significant trend towards improved child mortality is consistent with contemporary estimates that were made at a local level in Tanzania and with studies based on parish records. This trend is contextualised against a historical analysis of the proximate determinants of child mortality. It is suggested that the improved availability and efficacy of pharmaceuticals in the 1950s was the key factor in prompting the downturn. Comparing the trend in child mortality with data from other sub-Saharan African countries raises questions about the relative efficacy of public health interventions in the interwar period under different colonial regimes.