The current experiments systematically examined semantic content integration as a mechanism for explaining source inattention and forgetting when reading-to-remember multiple texts. For all 3 experiments, degree of semantic overlap was manipulated amongst messages provided by various information sources. In Experiment 1, readers' source recognition was significantly poorer when the sources presented semantically-congruent compared to semantically-distinct messages. Experiment 2 replicated the findings, despite half of the participants receiving a pre-reading warning. Experiment 3 extended the examination to include longer argument-based texts; readers additionally wrote a comprehensive essay on the topic. The results indicated longer reading times and better recall memory for the claims and evidence statements from semantically-congruent compared to semantically-distinct texts, while still reproducing the poorer source recognition effects of Experiments 1 and 2. We discuss implications for contemporary accounts of multiple text comprehension as well as directions for future research.