Research indicates that mathematical fact retrieval difficulty is one of the main manifestations of mathematics learning disabilities. Traditionally, interventions for fact retrieval difficulties of students with math learning disabilities have placed focus on mastery of algorithms and repeated practice; however, such instructions failed to address these students’ underlying cognitive deficits and can be insufficient and may hinder students’ development of mathematical competence in the long run. Empirical studies in neuropsychology have provided evidence that lack of number sense can casually affect individual’s math performance. Recently, researchers in the field of math learning disabilities have also pointed out that lack of number sense may causally affect fact retrieval performance of students with math learning disabilities. However, to date, no empirical study has been conducted to examine the effectiveness of number sense instruction on students with math learning disabilities. Therefore, the purposes of this study were: (1) to investigate the effectiveness of number sense instruction on fact retrieval performance of students with math learning disabilities; (2) to examine whether students with math learning disabilities who receive instruction in number sense can generalize their understanding of number and number systems to untaught math skills. A single subject multiple treatments (A-B-A-C and A-C-A-B) design across subjects was used to examine the effect of number sense instruction versus instruction using repeated practice on six students who showed early signs of math learning disabilities. Students were divided into two groups. Students in group 1 received number sense instruction followed by repeated practice; students in group 2 received repeated practice followed by number sense instruction. The results showed that students who received repeated practice followed by number sense instruction had better initial performance on fact retrieval. However, students who received number sense instruction followed by repeated practice had higher performance on fact retrieval toward the end and could generalize what they learned to more novel tasks, such as solving word problems.