Purpose: Our goal was to compare statistical learning abilities between preschoolers with developmental language disorder (DLD) and peers with typical development (TD) by assessing their learning of two artificial grammars. Method: Four- and 5-year-olds with and without DLD were compared on their statistical learning ability using two artificial grammars. After learning an "aX" grammar, participants learned a relatively more complex "abX" grammar with a nonadjacent relationship between "a" and "X." Participants were tested on their generalization of the grammatical pattern to new sequences with novel X elements that conformed to ("aX," "abX") or violated ("Xa," "baX") the grammars. Results: Results revealed an interaction between age and language group. Four-year-olds with and without DLD performed equivalently on the "aX" and "abX" grammar tests, and neither of the 4-year-old groups' accuracy scores exceeded chance. In contrast, among 5-year-olds, TD participants scored significantly higher on "aX" tests compared to participants with DLD, but the groups' "abX" scores did not differ. Five-year-old participants with DLD did not exceed chance on any test, whereas 5-year-old TD participants' scores exceeded chance on all grammar learning outcomes. Regression analyses indicated that "aX" performance positively predicted learning outcomes on the subsequent "abX" grammar for TD participants. Conclusion: These results indicate that preschool-age participants with DLD show deficits relative to typical peers in statistical learning, but group differences vary with participant age and type of grammatical structure being tested.