Resources that can optimize survival are distributed unevenly across landscapes. I assumed that the distribution of these resources is reflected by landscape heterogeneity and that selection for areas with relatively high heterogeneity reflects on selection for resources. I therefore hypothesized that landscape heterogeneity determines variability in elephant home range location and size. I evaluated home range sizes for elephants living in the Etosha National Park (n = 6), Khaudum Game Reserve (n = 6) and Ngamiland District 11 (n = 4) during two wet and two dry seasons. I used raster grid landscape maps, which were based on structural classes, to superimpose elephant home ranges and to generate randomly located ranges. I then used the FRAGSTATS programme to calculate five landscape metrics that measure aspects of heterogeneity within elephant and randomly located ranges. I compared landscape heterogeneity of the three study areas using the landscape metrics calculated for the randomly located ranges. Assuming that higher values of Patch density, Lanscape shape index and Shannon diversity index, and lower values of Largest patch index and Contagion implies selection for heterogeneity, I qualitatively compared the distribution and the mean of the landscape metric values for the elephant home ranges with those for the randomly located ranges. The influence of season on selection for the landscape metrics was also evaluated. I searched for relationships between home range size and landscape metrics for the three study areas and searched for a possible relationship between home range size and water point density for elephants in Etosha and Khaudum. Khaudum was consistently more heterogeneous than Etosha and NG11, while the latter was the least heterogeneous. Within these study areas, at least some of the landscape metrics may explain the location of elephant home ranges. The consistent apparent selection for areas with relatively higher values of Landscape shape index and Shannon diversity index and lower values of Largest patch index and Contagion across both wet and/or dry seasons by elephants in Etosha and NG11 suggests that landscape heterogeneity can explain landscape selection by elephants. This did not hold for elephants in Khaudum where elephants inconsistently selected for Largest patch index and Landscape shape index. This study further suggested that landscape heterogeneity did not determine home range size. However, water was a strong determinant of home range size and may therefore explain the lack of selection for heterogeneity in Khaudum. Heterogeneity is an inherent characteristic of landscapes and seems to reflect on the availability of resources that may subsequently influence the way elephants use space. My study supported the concept that increasing space, while promoting heterogeneity for elephants may enhance range expansion. The management of elephants should therefore be directed at optimising the availability of heterogeneous landscapes when setting land aside for their conservation. Management also should reconsider water distribution policies since the availability of water influences landscape use and conceivably potential impact on vegetation.