This study concerns the feasibility of a Markov chain model for projecting housing values and racial mixes. Such projections could be used in planning the layout of school districts to achieve desired levels of socioeconomic heterogeneity. Based upon the concepts and assumptions underlying a Markov chain model, it is concluded that such a model is not by itself an adequate tool for forecasting future changes in the distribution of housing values. However, such a technique is found to be a powerful descriptive device which provides insights essential to the construction of a deterministic model not having the Markovian weaknesses. Results obtained in test runs of the model suggest that it is worthwhile to pursue the technique in a larger context. (TT)