In this paper, a two-dimensional (2-D) spectral estimation is presented. The technique is based on impulse responses carried on the rows and columns of a data matrix. The 2-D frequencies that give the spectrum of a protein molecule in the image plane are obtained from the eigenvalues of open-loop matrices derived from the impulse responses. A modal decomposition of one of the open-loop matrices is used to pair the frequencies without effort. The amplitudes associated with the 2-D frequencies are computed from least-squares of the eigenvalues of the open-loop matrices onto the data. Performance of the technique is tested by computer simulation; effectiveness is confirmed, and results are presented for 2-D nuclear magnetic resonance (NMR) data.