Neurological soft signs (NSSs) are subtle motor and sensory deficits, and are associated with poor cognitive abilities. Although cognitive ability has been found to be a significant predictor for academic performance in children, it remains unclear whether NSSs could contribute to academic abilities such as mathematical skills, and its contribution varies according to grade level. Therefore, in this cross-sectional study, we examined the relationships between NSSs and different mathematical skills (calculation fluency, numerical operations, and mathematical problem-solving) in 105 Chinese children (Mean age = 7.76 years, SD age = 0.67 years; 52 from second grade, 53 from third grade; 56 boys and 49 girls) recruited from a primary public school located in Shanghai. The results of regression analyses revealed that NSSs significantly predicted calculation fluency (β = -.32, p < .050), numerical operations (β = -.38, p < .050), and mathematical problem-solving (β = -0.40, p < .010) in second but not third grade, even controlling for cognitive processes. Our results implicate that NSSs could be a potential predictor for mathematical skills in the early years of primary school.