Understanding heat, moisture and mass transport during the roasting of a coffee bean is essential to identifying how the colour and flavours are produced. This paper first considers a slightly simplified version of an existing heat and moisture transport model proposed by Fabbri et al. (2011), and we show that this model can be fitted well to data for the moisture content of a coffee bean but has some stability issues and lacks some important physical mechanisms. Building on these ideas, a new model is derived from conservation equations. This model is then simplified; in particular, issues of CO2 production are neglected as there is currently insufficient experimental data to fit parameters. This new model is fitted to the same experimental data as presented by Fabbri et al. The new model predicts significantly different internal structure and behaviour of the moisture than the existing model, while both show qualitatively similar average behaviour. This is due to the fact that our model tracks local, rather than bulk, quantities. One benefit to this new model is that it accurately predicts the existence of a sharp drying front, which partitions the bean into an outer dry region and an inner moist region. A detailed comparison of the two models is provided, in order to cast light on the relative importance of various heat and mass transfer mechanisms inherent in coffee bean roasting. (C) 2016 Elsevier Ltd. All rights reserved.