There is little order or formal direction to current cost modelling research. This paper proposes a basic classification system with which both to set out the overall topology of cost modelling as a distinct subject, and to provide the much needed points of reference on which individual research contributions can be located. The taxonomy is used to review over 50 reported cost models. The analysis of this review highlights that current emphasis lies with non-specific, macro, price models which use abstract units of measurement, and are intended to be used at the sketch design stage. These models generally use simulation techniques based on functional dependencies, with implicit assumptions and a deterministic outcome. On current trends, the preferred models of the future will use as-built units of measurement, and make considerably more use of stochastic techniques. One criteria for which there appears far too little concern, is in making the assumptions within the model more explicit.