Solution for any problem calls for its proper assessment and estimate. Polices, both macro and micro, for providing safety-nets for the poor in developing countries like India are often formulated with inadequate appreciation of empirical and methodological aspects. As a result factually incorrect assessments of the problem and its magnitude are made, which lead to wrong choice and design of instruments and policies. This issue is examined with reference to India. There is a consensus that incidence of poverty with reference to the calorie intake criterion has declined since the mid-1970s to about 35 per cent of the population. The study examines how far this consensus is valid? How far the database is suitable for such assessments? What are its implications for the observed trends in poverty estimates in the context of dynamic structural changes in the rural economy? The study concludes that the estimates do not show a real reduction in poverty but only a reduction in overestimation of poverty for the initial years followed by its underestimation for the later years. Even today about 75 per cent of the population is calorie deficient largely due to income poverty. This calls into question the choice of policy measures to reduce fiscal deficit by restricting welfare programmes and poverty alleviation measures only to the population currently believed to be poor.