Two geographically distinct silcrete associations are present in southern Australia, inland and eastern; these were sampled in central South Australia and central Victoria, respectively. At each site, both silicified and immediately adjacent unsilicified parent material were collected. Analytical data from these pairs were used to construct isocons, assuming Zr immobility, and to calculate the volume change and amount of silica introduced during silicification. These results, together with whole-rock oxygen-isotope compositions, were used to determine the delta 18 O of the introduced silica. The results show that the eastern silcretes in central Victoria are probably linked genetically to the associated basalts, weathering of which supplied the introduced silica. This conclusion is based on the close spatial connection between the two, as well as the substantial amount of introduced silica in the silcretes (greater than in the inland silcretes), resulting in volume increases in some eastern silcretes. Oxygen isotopic calculations for the silcretes indicate that the silica precipitated from groundwaters at temperatures slightly higher than present conditions. Silcrete formation apparently occurred during the Miocene and Pliocene (basalts in Victoria younger than Pliocene lack associated silcrete) and may reflect the much wetter climate in southeastern Australia at that time. The inland silcretes of central South Australia can be divided into pedogenic (the most common) and groundwater varieties. The pedogenic silcretes, which show typical soil features like columnar and nodular textures, contain moderate amounts of introduced silica that precipitated by evaporation from saline groundwaters. For the groundwater silcretes, which have massive textures and formed at or close to the water table, insufficient data are available to determine the mode of formation. The inland pedogenic silcretes have probably been forming from the Eocene-Miocene to the present, implying that conditions of seasonally high evaporation have occurred in central Australia during this time period. Thus silcrete formation depends on a complex interplay between climate and silica supply, and it is impossible to generalize that the presence of silcrete is indicative of a particular climate. Likewise, the elemental composition of silcretes, particularly Ti content, is not necessarily of climatic significance. Nevertheless, detailed geochemical and oxygen isotopic studies of a silcrete and its parent material can elucidate the mechanisms of silcrete formation, and if evaporation is indicated as a major factor in silcrete formation, then the climate at the time was likely to have been at least seasonally arid.