A new two-dimensional spectral estimation procedure, termed semicausal, is applicable to analysis of ocean gravity waves. Spectral estimates of both reference functions and actual synthetic aperture radar data of ocean waves have been generated using semicausal techniques and compared to fast Fourier transform estimates of identical data sets. The semicausal method can successfully generate spectral estimates of truncated data sets and data sets with two closely spaced frequency components. The semicausal estimate is sensitive to the autoregressive order and exhibits spectral splitting in some cases. Its noise sensitivity is similar to that of the Fourier transform.