Abstract. In this paper we provide a theoretical framework for estimating the size of the intersection join between two line segment datasets (e.g., roads, railways, utilities). For real datasets, it has been pointed out that the line segment lengths and slopes are distributed according to specific mathematical laws [14]. Starting from this result, we show how to predict the size of the intersection join between two line segment datasets. We evaluate our formula through several experimentations, showing that the estimation is accurate, as compared to that obtained by using a naive uniform model.