Clustering of bus flow easily occurs due to the influence of bus stop. A new cell automaton model which considers the length and distribution of bus stations is proposed in this paper. Numerical simulation shows that the model can explicitly depict the space-time trajectories of bus movement along a bus line, reproduce the clustering phenomenon of bus flow, and quantitatively evaluate the impact of bus station distribution on bus average speed. The study provides insight on formation mechanism of clustering and may contribute to the optimization of locating bus stations.