In 2001 the Minimum Income Guarantee for UK pensioners was reformed, changing the structure and level of benefits. We evaluate the behavioural response to this reform, using nonparametric analysis comparing a sample of pensioners interviewed before and another interviewed after the reform, matching their simulated pre- and post-reform entitlements and other characteristics. We compare the results with conventional parametric methods and also "ex ante" matching, and we consider the effect of measurement error in simulated entitlements. The response of take-up to the reform is found to be significant and positive, with evidence of larger impacts from the nonparametric analysis. Copyright (c) The London School of Economics and Political Science 2008.