Gesture and speech are assumed to form an integrated system during language production. Based on this view, we propose the integrated‐systems hypothesis, which explains two ways in which gesture and speech are integrated—through mutual and obligatory interactions—in language comprehension. Experiment 1 presented participants with action primes (e.g., someone chopping vegetables) and bimodal speech and gesture targets. Participants related primes to targets more quickly and accurately when they contained congruent information (speech: “chop”; gesture: chop) than when they contained incongruent information (speech: “chop”; gesture: twist). Moreover, the strength of the incongruence affected processing, with fewer errors for weak incongruities (speech: “chop”; gesture: cut) than for strong incongruities (speech: “chop”; gesture: twist). Crucial for the integrated‐systems hypothesis, this influence was bidirectional. Experiment 2 demonstrated that gesture’s influence on speech was obligatory. The results confirm the integrated‐systems hypothesis and demonstrate that gesture and speech form an integrated system in language comprehension.