Transfer opportunities play an important role in modeling public transport networks. Finding the optimal transfer locations is essential for providing valuable passenger information in order to support journey planning. Furthermore, using this information makes possible to accelerate general-purpose shortest path algorithms, making transit assignments more efficient. However, determining optimal transfer opportunities in a dense network is not an easy task. This paper presents the principles of a possible solution, and a new algorithm based on them, illustrated with actual examples. The algorithm has a polynomial complexity; therefore it can be used easily in practice.