A raft of recent studies has highlighted a major decline in large mammal populations in many of Africa’s protected areas. A recent continent-wide assessment represented a major step forward also in terms of quantifying the decline on a regional basis, but fell short in its sampling and analysis. In this paper, a way out of the “black box” of large mammal declines in Africa’s protected areas is formulated, with the aim of assisting in the preparation of further assessments in the future. First, large mammal assessments are categorized, highlighting the importance of using and sometimes juxtaposing peer-reviewed sources. The importance of the length of time series of large mammal counts is stressed, allowing one to distinguish between natural variation (especially rainfall) and human-induced changes. Setting reference dates, such as 1970, often gives biased results, showing the need to interpret large mammal population assessments in a historic context as well. This holds true particularly for West-Central Africa, which has experienced a considerable decline in rainfall since 1970. Building on a framework that connects herbivore physiology and behavior with environmental gradients, examples are given to explain some striking observed changes. The declines in Africa’s protected areas are not limited to large mammals, but have also been observed for large birds of prey. Assessments of large mammal populations should be accompanied by the identification of proximate drivers of change, for which a framework is suggested in this paper. To conclude, some suggestions for countering the declines in large mammal populations are presented.