The fieldwork for the 61st National Sample Survey (NSS) has recently been completed and Indian poverty estimates for 2004/05 are currently being calculated. We revisit the debate over the 55th NSS round, which produced incomparable poverty estimates due to a change in the recall period used in the survey. The adjustment methods of Sundaram and Tendulkar (2003a; 2003b); Deaton (2003a and Drèze 2002); and Kijima and Lanjouw (2003) are analysed, and the underlying assumptions of each method are probed. We conclude that the use of the Employment/Unemployment surveys is not valid, reject the assumption of a stable Engel curve but cannot reject theassumption that returns to factors are constant. With the caveat that there has been no structural shift in the relationship between income and expenditure, we accept Kijima and Lanjouw’s adjustment method. This implies that poverty reduction during the 1990s was considerably slower than prior to the 1991 economic reforms.