Many aspects of children’s language environment (including input word-type and -token counts and reading) and demographic variables (including maternal education and child gender) are correlated with early language abilities. Here we ask which aspects of early language input relate to later language skills, controlling for demographic influences. We analyzed children’s language input and production in infancy (age 0;6—1;5) and their language skills two years later (age 3;6; n=37/44 completed follow-up). First, we provide the first longitudinal validation of the Quick Interactive Language Screener (QUILS), a new preschool language measure. We find that QUILS scores are correlated with parent-reported infant vocabulary (CDI), observed infant vocabulary (from at-home recordings), and concurrent vocabulary scores (PVT; all ps<.05). We also ran several models predicting QUILS scores with measures of infant language input. The best model predicting preschool language with input measures included more noun tokens and relatively fewer nouns from TV & toys. Summarily, we find evidence that language input and production in infancy are tied to language abilities even years down the road.