This paper seeks to develop a robust set of tools for estimating the stochastic properties of sea ice drafts. Five parameters are estimated, along with their formal uncertainties: mean draft, rms topographic variation, characteristic length, fractal dimension, and normalized skewness. These provide an objective basis for terrain classification. A recent 10-km by 10-km upward looking submarine sonar survey serves as a test case. Data profiles are compared with synthetic profiles generated from the statistical parameters estimated from the data. These provide a means of qualtatively assessing the success of the stochastic model at characterizing the principal morphological properties. In general, the comparisons are good, providing confidence in the applicability and versatility of the stochastic model. A few exceptions are notable, but these also provide value insight into our understanding of the morphology of sea ice draft. ¿American Geophysical Union 1995