Purpose: The authors of this study investigated the validity of tense and agreement productivity (TAP) scoring in diverse sentence frames obtained during conversational language sampling as an alternative measure of finiteness for use with young children. Method: Longitudinal language samples were used to model TAP growth from 21 to 30 months of age for 37 typically developing toddlers. Empirical Bayes (EB) linear and quadratic growth coefficients and child sex were then used to predict elicited grammar composite scores on the Test of Early Grammatical Impairment (TEGI; Rice & Wexler, 2001) at 36 months. Results: A random-effects quadratic model with no intercept best characterized TAP growth, replicating the findings of Rispoli, Hadley, and Holt (2009) . The combined regression model was significant, with the 3 variables accounting for 55.5% of the variance in the TEGI composite scores. Conclusion: These findings establish TAP growth as a valid metric of finiteness in the 3rd year of life. Developmental and theoretical implications are discussed.