Rift Valley fever (RVF) is a zoonotic and vector-borne disease, mainly present in Africa, which represents a threat to human health, animal health and production. South Africa has experienced three major RVF epidemics (1950-51, 1973-75 and 2008-11). Due to data scarcity, no previous study has quantified risk factors associated with RVF epidemics in animals in South Africa. Using the 2008-11 epidemic datasets, a retrospective longitudinal study was conducted to identify and quantify spatial and temporal environmental factors associated with RVF incidence. Cox regressions with a Besag model to account for the spatial effects were fitted to the data. Coefficients were estimated by Bayesian inference using integrated nested Laplace approximation. An increase in vegetation density was the most important risk factor until 2010. In 2010, increased temperature was the major risk factor. In 2011, after the large 2010 epidemic wave, these associations were reversed, potentially confounded by immunity in animals, probably resulting from earlier infection and vaccination. Both vegetation density and temperature should be considered together in the development of risk management strategies. However, the crucial need for improved access to data on population at risk, animal movements and vaccine use is highlighted to improve model predictions.