Mathematics achievement of Canadian elementary students has recently declined on provincial standardized testing. Despite a current emphasis on accountability-driven education, limited research has acknowledged the relation between heterogeneous student need and patterns of student achievement, specifically in regard to students with individual education plans. Therefore, the objective of the present study was to identify latent profiles of elementary student mathematics achievement on Ontario's Education Quality and Accountability Office test and to determine factors that predict membership to these subpopulations. Latent class modeling revealed two profiles of inclining or declining student mathematics achievement, with individual education plans and testing accommodations significantly predicting membership to a latent profile of declining mathematics achievement, as determined by a binary logistic regression. Findings are discussed relative to their implications for educational practice by questioning the effectiveness of current testing accommodations and advocating for early identification and intervention to support mathematics learning.