Using 1.2.3 survey data on the Democratic Republic of Congo, we highlight different sectors in the labor market, with ‘higher paid’ sectors that are largely formal and ‘lower paid’ sectors that are largely informal. Based on a linear regression model, we report significant heterogeneity in earnings across sectors, which remain after controlling for aspects of human capital. We use a multinomial logit model to identify sector choice and show, allowing for selection, how returns to human capital differ across these sectors. We find that returns to basic education are important in largely informal sectors and that tertiary education is very important to access the higher paid sectors, but less important than in the lower paid sectors in increasing earnings once there. We then extend this analysis using quantile regression to show how returns differ across the distribution within sectors. Finally, we decompose the earnings gap across sectors and check how characteristics and return on these characteristics affect the cross-sector earnings gap. The earnings gap decomposition shows that workers of the lower paid sectors earn less, not only because they are less skill-endowed but also because they earn lower returns on their skills.