Researchers agree that both domain-general skills and domain-specific skills contribute to mathematics knowledge, but questions arise as to which skills can and should be trained to improve children's learning outcomes. In this article, we synthesize research on training three domain-general constructs in early childhood (patterning skills, working memory, and spatial reasoning) and their causal links to mathematics knowledge. The results are clear: Practice with these domain-general tasks is unlikely to transfer to higher scores on measures of mathematics knowledge, especially for preschool-aged children. Based on this evidence, we argue against using isolated domain-general training to enhance math knowledge in early childhood. We offer recommendations for researchers and practitioners to optimize best practices in this area.