A method is presented for modelling droplet size distributions within condensing steam flows. Polydispersed droplet spectra are treated by modelling moments of the size distribution, rather than the more traditional approach of discretising the continuous range of radii. Attention is focussed on wet-steam flow in turbines for which accurate prediction of droplet spectra is an essential step towards computing thermodynamic wetness losses and other two-phase effects. The method has been validated by comparison with independent calculations for both primary and secondary nucleation. Excellent agreement was obtained throughout, but with substantial reduction in complexity and computer usage. The moment equations have been cast in both Lagrangian and Eulerian forms and are well suited for incorporation into computational fluid dynamic codes. (C) 2000 Elsevier Science Ltd. All rights reserved.