Large herbivores such as elephants (Loxodonta africana) apparently have a negative impact on woody vegetation at moderate to high population densities. The confounding effects that fire, drought, and management history have may complicate assignment of such impacts to herbivory. We reviewed 238 studies published over 45 years and conducted a meta-analysis based on 21 studies that provided sufficient information on response of woody vegetation to elephants. We considered size and duration of studies, elephant densities, rainfall, fences, and study outcomes in our analysis. We detected a disproportionate citation of 20 published studies in our database, 15 of which concluded that woody vegetation responded negatively to elephants. Our analysis showed that high elephant densities had a negative effect on woody vegetation but that rainfall and presence of fences influenced these effects. In arid savannas, woody vegetation always responded negatively to elephants. In transitional savannas, an increase in elephant densities did not influence woody vegetation response. In mesic savannas, negative responses of woody vegetation increased when elephants occurred at higher densities, whereas elephants confined by fences also had more negative effects on woody plants than elephants that were not confined. Our analysis suggested that rainfall and fences influenced elephant density related impact and that research results were often site-specific. Local environmental conditions and site-specific objectives should be considered when developing management actions to curb elephant impacts on woody vegetation.