A technique for estimating the ocean surface roughness probability density function from satellite altimeter data is presented. Results from the application of the technique to Geos 3 altimeter data demonstrate its ability to detect both large-scale and small-scale structural deviations from the Gaussian distribution. Knowledge of the surface roughness probability density enables a direct computation of significant wave height. Significant wave height computations using the estimated quantities have been found to be in general agreement with other approaches. Density function estimation is important to precision altimeter design and has possible implications in oceanographic research.