Purpose: This study examined whether the Predictive Early Assessment of Reading and Language (PEARL), a dynamic assessment of narratives that measures language comprehension and production, accurately classifies Navajo preschoolers with typically developing (TD) language or with language impairment (LI). Method: Ninety 4- and 5-year-old Navajo preschoolers were identified as having LI or are TD (n = 45 each) via a 5-measure battery: parent report, teacher report, English narrative, independent educational plan, and the Clinical Evaluation of Language Fundamental Preschool--Second Edition (Wiig, Secord, & Semel, 2004). Children completed a PEARL pretest, a narrative mediation phase providing principles of narrative structure, and a PEARL posttest. A modifiability score reflected responsiveness to mediation. Results: The PEARL pretest and posttest each distinguished children with LI versus TD children with 89% accuracy; modifiability scores identified children with 100% accuracy. The PEARL story grammar subtest at pretest and posttest best distinguished LI versus TD. A revised cutoff score on the PEARL pretest decreased the diagnosis of TD children as having LI; the standard PEARL posttest cutoff was retained. Conclusion: The PEARL is a promising assessment for accurately differentiating Navajo preschool children with LI from those with TD language, particularly with a revised pretest cutoff score.