This study is purposed to investigate gender gaps in mathematics within (boys vs. girls in each ethnic group) and between four major ethnic groups (boys vs. boys and girls vs. girls between ethnic groups) using a nationally representative data, ECLS-K (Early Childhood Longitudinal Study-K Cohort). The sample includes 9,824 Whites, 2,469 African American, 3,015 Hispanic, and 1,100 Asian students. Two Analyses of Variance (ANOVA) models based on gender were conducted to assess the statistical significance of differences among ethnic groups and Latent Growth Curve (LGC) models were used to measure growth functions at each point of times. According to the results, gender gaps in mathematics performance were statistically significant and varied by ethnicity at different points in time. White male students consistently showed significantly higher scores on mathematics than White female students from kindergarten through the third grade. Among other ethnic groups (African Americans, Hispanics and Asian Americans) gender gaps in mathematics were not as consistent. However, regardless of student ethnic backgrounds, a gender gap favoring male students became statistically significant at the third grade. White male students also scored significantly higher than male students of other ethnic groups from kindergarten through the third grade, and White female students scored higher than girls from other ethnic groups.