The aim of the study was to determine whether memory performance predicted mathematics achievement in primary school 3rd graders. The sample consisted of 144 3rd grade students selected via convenience sampling method, and drawn from two public and one private schools in Istanbul, Turkey. The three subtests of the Wide Range Assessment of Memory and Learning 2 (WRAML2), namely Symbolic Working Memory subtest, Design Memory subtest, and Finger/Windows subtest were used to assess memory performance. The arithmetic subtest of the Wide Range Achievement Test III (WRAT-III) was used to assess mathematics performance. Prior to conducting a regression analysis measuring the predictive role of memory performance on mathematics achievement, the internal consistency and test-retest reliability of the subtests were computed. Internal consistency measured by Cronbach's alpha and test-retest findings provided evidence for reliability of all subtests. The Symbolic Working Memory subtest and the Design Memory subtests determined mathematical achievement in a linear regression analysis. The results revealed that WM performance measured by the Symbolic Working Memory subtest and visual memory measured by the Design Memory Subtest of WRAML2 determined arithmetic achievement, whereas the Finger/Windows Subtest measuring sequential memory/directed attention didn't predict arithmetic achievement.