The author assesses the performance of Latvias system of social transfers, in three ways: First, he analyzes the incidence (who receives transfers) of pensions, family allowances, unemployment benefits, and social assistance. Per capita analysis shows pensions tending to be pro-rich, and families allowances pro-poor (a finding typical in poverty analyses). Introducing an equivalence scale alters the results and shows all individual cash transfers performing about the same: mildly pro-poor. Next, he examines the performance of social assistance, which is, by definition, directed to the poor. He shows that Latvia's current system is concentrated - meaning that social assistance is disbursed to few households (only 1.5 percent of all households receive it) but among those that do receive it, it represents a relatively high share (20 percent) of income. Households that are systematically discriminated against in the allocation of social assistance are urban households living outside the capital (Riga) and those headed by male adults. Third, he looks at the regional allocation of social assistance. The results confirm earlier findings of large horizontal inequalities - that people with the same income from different parts of the country are treated unequally, because the existing system is based on local financing of social assistance.