Gesture is an important precursor of children’s early language development, for example, in the transition to multiword speech and as a predictor of later language abilities. However, it is unclear whether gestural input can influence children’s comprehension of complex grammatical constructions. In Study 1, 3- (M = 3 years 5 months) and 4-year-old (M = 4 years 6 months) children witnessed 2-participant actions described using the infrequent object-cleft-construction (OCC; It was the dog that the cat chased). Half saw an experimenter accompanying her descriptions with gestures representing the 2 participants and indicating the direction of action; the remaining children did not witness gesture. Children who witnessed gestures showed better comprehension of the OCC than those who did not witness gestures, both in and beyond the immediate physical context, but this benefit was restricted to the oldest 4-year-olds. In Study 2, a further group of older 4-year-old children (M = 4 years 7 months) witnessed the same 2-participant actions described by an experimenter and accompanied by gestures, but the gesture represented only the 2 participants and not the direction of the action. Again, a benefit of gesture was observed on subsequent comprehension of the OCC. We interpret these findings as demonstrating that representational cospeech gestures can help children comprehend complex linguistic structures by highlighting the roles played by the participants in the event.