Growing trend of using spatial information in various domains has increased the need for spatial data analysis. As spatial data analysis involves the study of interaction between spatial objects, Probabilistic Relational Models (PRMs) can be a good choice for modeling probabilistic dependencies between such objects. However, standard PRMs do not support spatial objects. Here, we present a general solution for incorporating spatial information into PRMs. We also explain how our model can be learned from data and discuss on the possibility of its extension to support spatial autocorrelation.