The aim of the study is to compare longitudinal patterns from Mathematics and Reading data from the direct child assessment of Early Child Longitudinal Study, Kindergarten (ECLS-K, US Department of Education, National Center for Education Statistics 2006), utilizing Profile Analysis via Multidimensional Scaling (PAMS). PAMS has been used initially to discover profile patterns in crosssectional data, and further applied to uncover longitudinal patterns by considering each time point as a coordinate of longitudinal patterns. The ECLS-K data analyzed here included longitudinal information about student achievement. The current study applied longitudinal PAMS to the data and examined how much the longitudinal patterns predict the fifth-grade achievement scores. Results showed that the longitudinal patterns that depicted the growing trend and the growing-decaying trend were significantly related to the fifth-grade achievement scores. Educational implications and discussions of longitudinal patterns were included. (Contains 3 tables and 10 figures.)