We provide a comparison of bat activity levels recorded during long-term acoustic monitoring through 188 microphones at pre-construction wind energy facility (WEF) sites in 12 South African ecoregions, and discuss the implications of the results for wind energy development. We summed all bat passes and detector hours over microphones, sites and years for each month, and fitted a negative binomial regression model with total bat passes as the response and ecoregion as the predictor. Overall, there was a significant effect of ecoregion on the number of bat passes per detector hour recorded near ground level, and in the turbine rotor sweep. Pairwise comparisons revealed that the sites in Maputaland coastal forests and woodlands, and KwaZulu-Natal-Cape coastal forests, were most distinct due to exceptionally high levels of recorded activity. As such, we strongly advise against WEF development in these ecoregions. In lowland fynbos and renosterveld, Limpopo lowveld, and Albany thickets, where intermediate to high bat activity was recorded, we recommend that the conditions of WEF-authorizations must include rigorous bat impact mitigation measures. For operational WEFs, our results provided valuable benchmark information for devising bat fatality thresholds that reflect the variation in bat activity across South Africa's diverse landscape.