The purpose of this study was to examine growth in early writing skills as assessed by curriculum-based measurements. The writing performance of 85 first-grade students was monitored across 12 weekly administrations of picture-word and sentence-copy measures using multiple scoring procedures. Data analysis was conducted using a model-building procedure in which random effect and quadratic effect parameters were systematically added to unconditional models to create best-fitting models for each scoring procedure and measure. Results suggested that random effect quadratic models fit picture-word tasks best, but that random effect linear models fit the data best for sentence-copy tasks. Additional analyses showed that a student's initial performance was moderately related to linear growth for a scoring metric that accounted for grammar, and that gender was related to initial performance but not growth. Limitations and implications for further use and research of curriculum-based measures for early writing are discussed.