This study examined gains in written language as assessed by targeted curriculum-based measures (CBMs), and explored how these gains were affected by moderator variables of specific cognitive functions and student subgroups. The sample included 68 second grade students who were at risk for writing disabilities. Handwritten compositions were collected throughout a written language intervention at baseline, sessions 3, 5, 10, 13, 15, 20, 22, and termination. Specific CBM variables included Total Number of Words Written, Words Spelled Correctly, Correct Word Sequences, and Percentage of Correct Word Sequences. Using latent growth curve analysis, models were estimated for each of the CBMs, but the data showed poor model fit. Latent class groupings using cognitive variables and student subgroups significantly moderated the growth rate for written language assessed by specific CBMs. Although these latter findings reflected potential moderators of change in written language, the lack of model fit raised questions around the use of these CBM variables in monitoring writing progress for second grade students at risk for writing disabilities. Findings from this investigation revealed the measurement complexities that likely remain hidden from teachers and other professionals engaged in routine progress monitoring using CBM variables. [This article is published in "Exceptionality" (EJ984498).]