This study examined the predictive validity of formative assessments embedded in a Tier 2 intervention curriculum for kindergarten students identified as at risk for reading difficulty. We examined when (i.e., months during the school year) measures could predict reading outcomes gathered at the end of kindergarten and whether the predictive validity of measures changed across the kindergarten year. Participants consisted of 137 kindergarten students whose reading development was assessed four times from October to February. Measures aligned with content taught in the curriculum and assessed a range of phonologic, alphabetic, and word-reading skills. Results from structural equation modeling indicate that 36.3% to 65.2% of the variance was explained on the latent decoding outcome and 62.0% to 86.8% on the latent phonological outcome across the four time points. Furthermore, the predictive validity of specific skills increased over the kindergarten year, with more complicated tasks (e.g., word segmentation) becoming more predictive at subsequent measurement occasions. Results suggest that curriculum-embedded measures may be viable tools for assessing and predicting reading performance.