An extension of a deep water hybrid parametric wave model has been developed for arbitrary water depths. The windsea is described by the JONSWAP parameter set and by a directional parameter representing the mean direction of the windsea spectrum. This mean direction has been re-derived based on more general energy flux arguments. The windsea spectrum in finite depth is assumed to have a similarity shape which is related to the deep water energy spectrum by multiplying with a depth dependent transformation factor. This factor affects the spectral shape of the hi gh-frequency part which is proportional to f-5 in deep water and to f-in shallow water. The directional dependence of the spectrum is assumed to be cos 2 (6-60), centered around the mean wave direction. The windsea model explicitly accounts for finite depth effects such as refraction, shoaling and dissipation by bottom friction.Based on an energy flux transport equation, the full set of prognostic parameters is derived including finite depth effects. The equations of the prognostic variables are solved on a finite difference grid by means of the Lax-Wendroff method. Swell is treated in a decoupled spectral fashion. For simplicity all swell characteristics are considered straight and effects of refraction are disregarded. However, shoaling and dissipation of energy by bottom friction are included.The model has been applied to a deep water and finite depth case study. The deep water hindcast was performed for an extratropical cyclone in the Gulf of Alaska. Predicted significant wave heights and zero-crossing wave periods are compared with measurements from six stations in the Gulf of Alaska. The finite depth wave conditions are hindcasted for a complex frontal system off the North Carolina coast during ARSLOE. Predicted spectral wave parameters such as significant wave height, spectral peak frequency and mean wave direction are compared with observations from several different measurement techniques. From the same data set, observed wave spectra are compared with hindcasted model spectra. A discussion of the results for both applications is presented.