People and elephants share landscapes throughout Mozambique. Here elephant conservation management focuses on protected areas but fails to address the conflict that exists between elephants and people. In this thesis I develop a landscape approach to conflict mitigation that is designed to accommodate the needs of people and of elephants in human-dominated landscapes. Mozambique faces a dilemma: politically it is required to reduce poverty while at the same time adhere to international agreements and requirements to protect biodiversity with relatively scarce financial resources. Reactive mitigation of human-elephant conflict (HEC) at the site-specific scale have proven to be costly and with low efficacy. A shift from reactive to proactive HEC mitigation approaches at the county-wide scale (e.g. a district level, the administrative planning body) may provide opportunities to reconcile such apparent contrasting requirements in Mozambique. The elephant population of Mozambique is fragmented and remnant sub-populations are limited to clusters of protected areas in a matrix of human-dominated landscapes. A meta-population perspective may accommodate this spatial structuring and allow for a conservation plan that ensures population persistence and moderate impacts with other species in the landscape. I assessed HEC throughout human-dominated landscapes of Mozambique to examine some assumptions associated with the landscape approach advocated here. I used spatially explicit human activity data, landscape features and elephant distribution at the grid cell of 25 km2 and at the district scale to test the practicality of landscape approaches to elephant conservation and mitigating HEC in the human-dominated landscapes of Mozambique. I then tested whether human activities have significant impacts on elephant numbers and distribution across Mozambique. Furthermore I tested if the costs and benefits of sharing space with elephants influenced HEC. Thereafter, I explored at the grain scale of 25 km2 if the degree of overlap between them on the use of resources can be used to predict the likelihood of HEC across the landscape. Direct and indirect human activities explained trends and rates of elephant population changes in Mozambique. Because most rural households of Mozambique rely on subsistence farming by extracting or cropping from the land, primarily for their own purposes, living close to elephant refuge areas represented a potential risk to humans. However, conflict with elephants does not centre on food security, but on lifestyle being affected by the presence of elephants, which itself was a function of human density. Rodents and insects are the primary agents responsible for food loss during food storage. HEC was not a function of elephant density – a combination of human density, percent cultivated area and human population growth rate best explained HEC incidences. Although at human densities beyond 60 people/km2, elephants disappeared, at low levels of land transformation and low human densities people and elephants co-existed, which may induce higher incidences of HEC. Proximity to roads and suitable land for agriculture were the best predictors for HEC in the rural areas of Mozambique. These results imply spatially driven causes of HEC. These findings supported assumptions that conservation landscapes embedded in different land uses that accommodate ecological needs of people and elephants as well as the likelihood of severity of HEC can ensure elephant conservation without forcing people into poverty. While reactive HEC mitigation actions at site-specific scales are attractive for local communities, proactive measures at the landscape scale may be more effective in the rural context of the distributional range of elephants in Mozambique. HEC can be mitigated proactively through an effective land-use planning that involves zonation and implementation. To address this I extrapolated the relevant findings from resources selection functions models at the 25 km2 grain scale for study locations to a country-wide scale and proposed a model of a likelihood of HEC. The country-wide HEC model yielded high predictive power and confirmed protected areas as sites of high elephant dependability. These models indicate focal areas for short to medium term reactive HEC mitigation measures and local community programs at specific site level. This dissertation suggests that human and elephant co-existence is possible in Mozambique. The apparent increase of HEC is not a function of numbers of elephants but of improper land use planning. In this thesis I argue in favour of a landscape approach to mitigate conflict between elephants and people. This approach should be considered in all national plans that aim to reduce conflict and enhance conservation.