A validated set of 8148 solid, subcooled liquid and liquid state vapour pressure data from literature for 1838 hydrocarbons and halogenated hydrocarbons was divided into a training set of 1200 compounds and a prediction set of 638 compounds. The training set was used to develop an artificial neural network for estimating vapour pressures at different temperatures. Input includes system temperature, 23 parameters calculable from chemical structure, and melting point for compounds solid at the temperature of interest. Training and prediction set give r² values of 0.995 and 0.990, and absolute average errors of 0.08 and 0.13 logarithmic units, respectively. The model can also be used to derive theoretically sound vapour pressure equations for individual compounds. In the appendix, a complete network description is provided to enable implementation and application of the derived model.