This article aims to make an empirical and theoretical contribution towards the creation of a continent‐wide dataset on African population extending into the pre‐1950 era. We investigate the reliability and the validity of the current population databases with the aim of working towards a consensus on the long‐term series of African total population with a reliable 1950 benchmark. The cases of Kenya, Nigeria, and Ghana are explored to show the uneven coverage of census taking in colonial and post‐colonial Africa and to demonstrate the need for an upward adjustment of the conventional 1950 benchmark. In addition, we discuss the advantages and disadvantages of Manning's approach of projecting population growth estimates backwards in time by adopting the available Indian census data as African ‘default growth rates’, and we propose an alternative approach by incorporating the demographic experiences of tropical land‐abundant countries in South‐East Asia.