Spatial data has been observed and recorded in various forms in numerous applications. Symbolic data analysis (SDA) is an emerging approach for aggregating large data sets into multi-valued forms, such as intervals, histograms, and lists. In this paper, we propose Bayesian methodologies for analyzing spatial interval-valued data (SIVD). We compare the prediction accuracy of the methods using proposed prediction performance metrics and examine prior sensitivity with different prior distributions. A real data set is used to illustrate the methods.