In the context of the controversy over the trends in poverty in India in the 1990s, this paper addresses the question of using the National Accounts Statistics (NAS) estimate of private final consumption expenditure (PFCE) in place of National Sample Survey (NSS) based estimate for calculating the proportion of the population below the poverty line or headcount ratio. A comparison is undertaken for the year 1993-94 between NAS estimate of PFCE and household consumer expenditure estimated from NSS disaggregated across item-groups of consumption and across selected fractile groups of the rural and urban population. Two alternative estimates from NAS - old series in NAS 1998 and the latest revision in NAS 1999 - are compared. Similarly, two alternative NSS estimates are considered - one directly available using a uniform 30-day reference or recall period and a synthetic one constructed to reflect the effect of using non-uniform reference period. The analysis of comparison of these four estimates suggests two major conclusions. One, the issue of accepting NAS estimate of PFCE as more correct and reliable than NSS estimate is far from settled. Two, the item-groups that accounted for a very large proportion of the aggregate discrepancy between NAS and NSS estimates had a much smaller budget share in the consumption basket of the bottom 30 per cent fractile group in the rural and urban areas, whereas in respect of item-groups which together accounted for over 75 per cent of the consumption of the bottom 30 per cent, the divergence between the two estimates was much smaller than on the average for all item groups and negative in some cases.