Using the 2000 introduction of means tested child benefits in Russia as a case study; this paper assesses the impact of a policy change from universal to means-tested child allowances in terms of targeting efficiency and poverty reduction. We focus on the core costs of targeting, namely leakage, exclusion errors and program costs, and related these to the benefits in terms of poverty reduction. We use the cross-section and panel dimensions of the Russia Longitudinal Monitoring Survey (RLMS) from 2000 to 2004 to analyse the impact of the reforms and to simulate the effects of various means-tested and universal child benefit schemes. Our analysis contributes to the academic debate in two ways. First, we compare means tested and universal schemes both in terms of targeting efficiency and poverty reduction effects. Second, we analyse the poverty reduction impact of these schemes also in terms of chronic poverty. We find that, since the reforms, more children receive benefits and that there is improved targeting of benefits to children living in low income households. Nevertheless, in 2004 one third to a quarter of the poor children does not receive benefits while about 50% of the benefits leaks to the three highest expenditure quintiles. The poverty reduction impact is modest and improved only marginally since the reforms. The simulations show that universal schemes achieve additional poverty reductions in all indicators because children who were by error excluded under the means tested scheme now receive a benefit.