PURPOSE: Preschoolers with language impairment have difficulties with both literal and inferential language, both of which are critical to later reading comprehension. Because these children are known to be at risk for later reading comprehension difficulties, it is important to design and test interventions that foster both literal and inferential language skills. Using a randomized pretest–posttest control group design, we investigated whether an 8-week, one-on-one book-sharing intervention would improve both the literal and inferential language skills of Head Start preschoolers with language impairments. METHOD: Thirty children were randomly assigned to either a control group that received no intervention or to a treatment group that received twice-weekly 15-min sessions in which adults read books and asked both literal and inferential questions about the books using scripts that were embedded throughout the text. Treatment and control groups were compared using pre- and posttest scores on 2 measures of literal and 1 measure of inferential language skill. RESULTS: Significant group differences, and medium to large effect sizes, were found between pre- and posttest scores for all 3 measures. CONCLUSIONS: These findings suggest that book sharing with embedded questions that target both literal and inferential language skills can result in gains on both types of language in this population. Future studies with larger number of children are needed to corroborate these findings.