Research Findings: This study builds on prior work related to the assessment of young dual language learners (DLLs). The purposes of the study were to (a) determine whether latent subgroups of preschool DLLs would replicate those found previously and (b) examine the validity of GOLD(R) by Teaching Strategies with empirically derived subgroups. Latent class analysis confirmed previous findings of 3 distinct latent subgroups of DLLs (bilingual children, emergent bilingual children, and heritage language speakers). Results of differential item functioning analysis showed that with few exceptions, GOLD items functioned similarly, which indicates that groups matched on ability were similar in their item scores. The item pertaining to using conventional grammar consistently favored non-DLLs over heritage language speakers. The item pertaining to name writing consistently favored DLLs as a single group, emergent bilingual children, and heritage language speakers. Practice or Policy: Study results provide further support for the heterogeneity of DLLs and the use of GOLD with DLL subgroups. This provides the field with an opportunity to better understand this special population of children and enables teachers to plan with greater precision experiences that contribute to their development and learning. (author abstract)