The SNARC effect reflected the fact that the participants’ reaction times(RTs) to small numbers were faster with left than right hand, whereas the converse for large numbers. Previous research suggested that the initial age of children show the SNARC effect is inconsistent，the mathematics education difference might be one of the reasons. In addition, previous studies focused on subjects in well-developed areas of education. But for China, there were regional differences in educational development. Especially in Guizhou, the education base and mathematics education of ethnic minority were far behind the developed areas. The SNARC effect of minority subjects in these areas should have different characteristics. Therefore, our study used parity decision task to test the developmental pattern of the SNARC effect of ethnic children and adolescents in Guizhou. In current research, 243 ethnic students of Guizhou were asked to judge parity between 8 Arabic numerals from 1 to 9 (except 5). To examine the SNARC effect, a repeated-measures ANOVA was conducted with response hand (left vs. right) and number size (1–4 as “small” vs. 6–9 as “large”) for each age group. And then, the regression method was used to further investigate the SNARC effect. We subtracted RTs or accuracy of the left-hand responses from those of the right-hand responses and regressed those differences on the magnitude of the numbers and obtained an unstandardized regression coef?cient for each participant. One-sample t test was used to test the significance of the mean unstandardized regression coef?cient of each group; Besides, to understand age differences, the data was analyzed in two ways: (1) The ANOVA approach was used to examine the three-way interaction (age by hand by number size); (2) One way ANOVA approach was used to examine whether there are grade differences in unstandardized regression coefficient. The results showed: (1) On the reaction times: 1) The interaction between number size and hand was significant for all groups (excluding grade 2);2) The significant T values which suggested a significant SNARC effect were existed in all stages of grade 3 to university;3) The three-way interaction was not significant, and the unstandardized regression coefficients of each grade group were also no significant difference; (2) On the accuracy: 1) The interaction between number size and hand was significant in grade 3 to grade 6. But, the significant T values were existed in grade 3 to grade 5，it suggests that the SNARC effect on accuracy were indeed existed in these three grades. 2) The three-way interaction was not significant, and the unstandardized regression coefficients of grade 3 to grade 5 were also no significant difference. These findings suggested that the ethnic children could exhibit SNARC effect from grade 3 which was later than urban children in developed areas. The size of the SNARC effect did not change with ages. Besides, grade 3 and grade 5 showed the same size SNARC effect on accuracy, the accuracy was too low (grade 2) or too high (Junior high school to University) to show the SNARC effect.