The issue of divergence between the estimates of private final consumption expenditure (PFCE) available from the National Accounts Statistics (NAS) and that from the quinquennial Consumption Expenditure Surveys (CES) of NSS has been under scrutiny for long. The former has not been used for official poverty estimation after 1993-94 due to the questionable reliability of PFCE in NAS. In this paper, it has been argued that before undertaking any cross-validation exercise, known and visible lacunae in the two data sets ought to be removed first as the difference, even at overall level, is too high. With this perspective in view, NSS data has been examined here and it has been found, inter alia, that there are ample evidence to favour direct methodological studies to improve NSS estimates.