The role of working memory on three- to seven-year-olds’ comprehension and production of temporal connectives was examined. The (target) sentences comprised two clauses joined by the temporal connectives before or after, presented in either a sentence-initial (e.g., before he played in the garden, he finished his homework), or a sentence-medial (e.g., he finished his homework before he played in the garden) position. Such sentence positions of the connective vary whether the order of mention of the events matches the temporal order, either chronological order (e.g., after he finished his homework, he played in the garden,) or reverse chronological order (e.g., he played in the garden after he finished his homework). To assess comprehension, children heard the sentence read aloud and were then presented with two pictures on a touch screen computer, representing the target object in each clause (e.g., garden, homework). Their task was to select the object representing the event that happened first in the sentence. To assess production, children were shown two videos and required to produce a sentence describing the two events. Memory was assessed using the digit span task from the Working Memory Test Battery for Children (Pickering & Gathercole, 2001). It was predicted that performance in the reverse-chronological sentences in both tasks would be poorer than for chronological order. Further, we tested whether performance in the reverse-chronological condition would be more strongly related to working memory because, for these sentences, the participant has to store and re-order the two clauses in order to construct an accurate situation model. Implications will be discussed in terms of how working memory capacity relates to children’s understanding of complex sentences and their construction of a coherent situation model.