The 55th round (1999-2000) of the NSS used a different methodology from all previous rounds and arrived at lower poverty estimates. The consensus from earlier NSS rounds, that poverty reduction had been set back during the 1990s, was challenged by this data. This was bolstered by some 'adjustments', which although agreeing that the 55th round had overestimated poverty reduction, claimed that the number of poor had nonetheless fallen by 30-45 million. However, a detailed re-examination shows that these 'adjustments' were not correct. The poverty ratio fell at most by 3 percentage points between 1993-94 and 1999-2000, and it is likely that the number of poor increased over this period. The main lesson is that poverty estimates are very sensitive to both survey design and post-survey analysis. The first of a two-part article.