To test whether a neonatal intensive care unit-based language curriculum for families with preterm infants enhances the language environment and postdischarge Bayley Scales of Infant and Toddler Development (BSID)-III language and cognitive scores. A randomized controlled trial was conducted with infants born at ≤32 weeks assigned to a parent-driven language intervention or health-safety lessons (controls). Recordings of adult word counts (AWC), conversational turns, and child vocalizations were captured at 32, 34, and 36weeks. Primary outcomes included 2-year BSID-III language and cognitive scores. We randomized 95 infants; 45 of the 48 intervention patients (94%) and 43 of the 47 controls (91%) with ≥2 recordings were analyzed. The intervention group had higher AWCs (rate ratio, 1.52; 95% CI, 1.05-2.19; P=.03) at 36weeks, increased their AWCs between all recordings, and had lower rates of 2-year receptive language scores <7 (10% vs 38%; P<.02). The intervention was associated with 80% decreased odds of a language composite score of <85 (aOR, 0.20; 95% CI, 0.05-0.78; P=.02), and 90% decreased odds of a receptive score of <7 (0.10; 95% CI, 0.02-0.46; P=.003); there was no association found with cognitive scores. Increases in AWC and conversational turns between 32 and 36weeks were independently associated with improved 2-year BSID-III language scores for both study groups. Short-term parent-driven language enrichment in the neonatal intensive care unit contributes to increased AWCs at 36weeks and improved 2-year language scores. In adjusted analyses, increases in conversational turns and AWCs at 36weeks were independently associated with improved language scores. This low-cost, easily implemented intervention can potentially help to mitigate speech delays among preterm infants. Registered with www. gov, NCT02528227.