1. The resolution of direct conflict between humans and elephants in Africa has become a serious local political issue in recent years, and a continental conservation problem. ‘Problem elephants’ damage crops, food stores and water sources, and sometimes threaten human life.
2. Eighty per cent of the African elephant's range lies outside formally protected areas, and inadequate management of human–elephant conflict is frequently a precursor to further decline in the numbers and distribution of elephants. Conflict appears to be increasing in an assortment of African ecosystems as the agricultural interface with elephant range expands.
3. The present study recorded incidents by problem elephants in small subdivisions of a 15 000 km2 elephant range. The level of problem elephant activity over 3 years showed huge variation and could not be explained by elephant density, proximity of a protected area, area of human settlement, human density or local rainfall.
4. It is proposed that the irregular and unpredictable nature of human–elephant conflict incidents in the study area mainly depended upon the behavioural ecology of individual elephant bulls.
5. This study proposes a statistic to quantify problem elephant activity in Africa which can be used to compare the intensity of problem incidents between different ecosystems at different times: ‘elephant incidents per square kilometre of human settlement area per year’. Spatial analyses of appropriate data at the human–elephant interface may yield a more predictive understanding of the conflict process.