Although termites are ecosystem engineers in tropical and sub-tropical environments, the study of termite ecology is often constrained by sampling difficulties and a lack of established sampling protocols, particularly for savannas. The efficiency and relevance of different methods along climatic gradients, even within a single biome, is largely unknown. Here, we compare the relative contribution of two commonly used sampling methods, cellulose baits and active searching transects, in quantifying savanna termite diversity along a rainfall gradient in South Africa; sampling was conducted during the wet season across four markedly different savanna types. We also assessed the usefulness of different forms of baiting techniques. The relative efficiency of sampling method varied with annual rainfall. In arid savannas, baiting was as effective as active searching transects at sampling termite diversity and we recommend the use of baiting rather due to it being less labor intensive. In savannas of moderately low to intermediate rainfall, baiting and transects sampled different termite species and so both are deemed necessary for an accurate assessment of termite diversity. In contrast, in wetter savannas transects gave a better assessment of diversity, with cellulose baits not contributing much to diversity assessment. The efficiency of baiting techniques differed across the rainfall gradient, with baits needing to be left in the field for a longer period in more arid savannas. We conclude that habitat type, even within a single biome, will determine the sampling method or methods necessary to quantify termite diversity accurately.