Young children's mathematics abilities may be divided between symbolic and non-symbolic skills. Lower performance of SES disadvantaged versus advantaged children has already been established in symbolic math. This study aimed to verify the effect of children's SES category on non-symbolic mathematical (numerical) performance. The main sample comprises 4,955 children from the French longitudinal study, ELFE, tested when they were in the nursery school (4- to 5-year-olds). The distinction between symbolic and non-symbolic math skills based on the specific math assessment items used in the present study was verified on a larger sample. The SES-related difference in non-symbolic math skills was then examined in the ELFE sample only. The children's performance in non-symbolic maths is significantly and almost as strongly correlated with their family's income and their mother's level of education as their performances in symbolic maths. Linear regression mixed-effects modelling shows that the score in non-symbolic maths (out of 100) of children from families with below median income is 3.8 points lower than that of their peers from families with above median income. Children from disadvantaged SES backgrounds perform significantly lower than those from advantaged backgrounds in non-symbolic maths. Even if non-symbolic math skills retain an educational interest, they should not reduce the importance of symbolic math skills in young children. © 2020 The British Psychological Society.