Space–time prism (STP) modeling offers a theoretical base for various person-based space–time accessibility measures. It has been challenging to construct STPs in multimodal transportation networks because of the vast multimodal trip chaining options between private vehicles and public transportation. This study suggests an effective method for narrowing down the trip chaining space. The method involves three incremental strategies to eliminate parking locations that contribute little to enlarging the STP. First, an upper bound potential path area (PPA) is constructed to remove all inferior parking locations. Second, the study area is partitioned and only complementary parking locations attractive to trip chaining are selected. Third, triangular inequalities are applied for pruning. The selected parking locations are used to construct a compact multimodal supernetwork encompassing one episode of activity participation. Consequently, the STP and PPA can be found by an existing two-stage bidirectional search method. Experimental tests are conducted to demonstrate the effectiveness of the method. It is found that 1–2 orders of speedups can be achieved with little compromising the accuracy of measuring space–time accessibility. The suggested method is applicable for measuring space–time accessibility in large-scale multimodal transportation networks.