The manipulation of fire is a technological act. The identification of the archaeological signatures of the controlled use of fire has important implications not only for the estimations of the origins and functions of the first fireplaces but also for our understanding of prehistoric technological development and resource use. At Riwi (Kimberley region, Western Australia), excavations over two field seasons have revealed a discontinuous occupation sequence over the past 45 ka, showing numerous, different combustion features interspersed within the deposit. Anthracological and micromorphological investigations at Riwi Cave indicate that the combustion features at the site can be categorised into three types: flat combustion features (type A), dug combustion features (type B) and thick accumulations of mixed combustion residues (type C). These provide evidence for two kinds of combustion practice: (i) fires lit directly on the ground and most likely not re-used and (ii) ground ovens, the latter appearing some 10,000 years after the first evidence for occupation of the site. A comparison of the wood species identified within these combustion features with those from equivalent scattered context levels, enables an exploration of the potential factors influencing wood selection and fire use through time at the site. A detailed understanding of the relationship between wood charcoal remains and archaeological context yields significant information on changes to environmental context and site occupation patterns over time.