Tephra, emplaced as a result of Pleistocene eruption of the Indonesian 'supervolcano' Toba, occurs at many localities in India. However, the ages of these deposits have hitherto been contentious; some workers have argued that these deposits mark the most recent eruption (eruption A, ca 75 ka), although at some sites they are stratigraphically associated with Acheulian (Lower Palaeolithic) artefacts. Careful examination of the geochemical composition of the tephras, which are composed predominantly of shards of rhyolitic glass, indicates that discrimination between the products of eruption A and eruption D (ca 790 ka) of Toba is difficult. Nonetheless, this comparison favours eruption D as the source of the tephra deposits at some sites in India, supporting the long-held view that the Lower Palaeolithic of India spans the late Early Pleistocene. In principle, these tephra deposits should be dateable using the K-Ar system; however, previous experience indicates contamination by a small proportion of ancient material, resulting in apparent ages that exceed the true ages of the tephras. We have established the optimum size-fraction in which the material from Toba is concentrated, 53-61 μm, and have considered possible origins for the observed contamination. We also demonstrate that Ar-Ar analysis of four out of five of our samples has yielded material with an apparent age similar to that expected for eruption D. These numerical ages, of 809 ± 51, 714 ± 62, 797 ± 45 and 827 ± 39 ka for the tephras at Morgaon, Bori, Gandhigram and Simbhora, provide a weighted mean age for this eruption of 799 ± 24 ka (plus-or-minus two standard deviations). However, these numerical ages are each derived from no more than 10-20% of the argon release in each sample, which is not ideal. Nonetheless, our results demonstrate that it is feasible, in principle, to date this difficult material using the Ar-Ar technique; future follow-up studies will therefore be able to refine our preparation and analysis procedures to better optimize the dating.