In developing the statistics of spatial data there is a need for methods in both the areas of data analysis and statistical z modeling. Here we analyze a data set of Sudden Infant Deaths, 1974 - 1978, in the counties of North Carolina, using a Markov-random-field approach to spatial modeling. We model the spatial trend with what we call large-scale-variation parameters, and the variance and spatial dependence with small-scale variation parameters. We show that a combination of resistant trend fitting to the data, and simple spatial auto Gaussian fitting to the residuals, is an effective way to analyze the (transformed) data.