Japan is a unique new immigrant destination because while the entire system is egalitarian compared to other societies, its educational system excludes immigrant children primarily due to the language barrier. Subsequently, we aim to scrutinize the following two questions of (1) whether co-ethnic neighborhood quality affects children’s opportunities for upper secondary education (high school enrolment) in Japan and, if so, (2) whether that effect varies across nationalities as an indicator of ethnicity. Grounded on the premise that the co-ethnic neighborhood effect depends on adaptation to the Japanese educational context, we utilize national census data in 2010 and employ a fixed-effect regression model to control the unobserved neighborhood effect. The random and fixed effect models support the positive effects of co-ethnic neighborhood qualities, including the percentages of highly educated residents, employed residents, and residents who have lived in the neighborhood since birth, even after considering children’s socioeconomic backgrounds and unobserved neighborhood heterogeneity. We also found that these effects vary across children’s nationalities.