This article estimates the proportion of grain "diverted" from the Public Distribution System (PDS) to the open market for several years in the past decade, by matching state offtake figures published by the government, with household purchase reported by the National Sample Survey (NSS). The limitations of this methodology are discussed, and alternative estimates are presented to crosscheck estimates obtained by using the conventional method. Though the alternative estimates are not very different, the discussion here indicates that estimates of diversion must be treated as an upper-bound. At the all-India level diversion of grain remains a serious issue, however, state-level trends
reveal interesting contrasts. Based on contrasting trends in monthly PDS purchase (per capita) and estimated diversion, states are categorized into three groups - "functioning", "reforming" and "languishing" states. The paper also discusses possible reasons for the improvement in the PDS in the reforming states.