We study the distributional impacts of carbon pricing policies using a multi-sector general equilibrium (GE) model with input-output linkages, heterogeneous agents and segmented labor markets. Households differ in their consumption patterns, labor types, and ownership of equity and capital. Pricing the carbon content of products affects households real income through an expenditure channel, according to the emissions intensity of their consumption baskets, and an earnings channel, as GE responses shift the relative demand for labor types, and returns on profits and rents. Calibrating the model with matched microdata for the Brazilian economy, we find regressive effects stemming from both channels. Ignoring the production networks and the gross complementarity between fuels, inputs and factors leads to a substantial underestimation of both aggregate and distributional effects. The incidence of the policy depends on how the revenue is recycled: expanding targeted social transfers fully offsets the regressive impact, whereas using the revenue to reduce preexisting consumption taxes improves efficiency but does not eliminate regressivity.