The provisional population totals of the 1971 census of India have shown population to be about 2.2 per cent or 14 million less than the official projection. A major explanatory factor appears to be an overestimation of the expectation of life during 1951-60 and the subsequent gains in mortality. A verification of the conjecture about decline in fertility having been greater than the assumed 5 per cent, will need intensive investigation when the 1971 age data become available. A likely important explanation might be a somewhat greater undercount than in the 1961 census. A comparison of the houselist estimates with the provisional population figures shows a surprising decline in the absolute population of three States of Uttar Pradesh, Bihar, and Rajasthan and relatively low growth rates during 1970-71 in several other States. There is an urgent need to launch a sample matching of the houselists and final census count to determine whether the houselist estimates overstated the number of normal residents or the census enumeration was affected by a significant undercount. The provisional figures have shown the population of Punjab to be 10 per cent smaller than was projected. On the other hand, in Tamil Nadu the growth rate has sharply accelerated. The underlying reasons need to be investigated with due attention to inter-district variations in growth rates, the likely contribution of migration to these variations, as well as the possibility that the observed growth rates during 1941-51 and 1951-61 might reflect differentials in the completeness of successive censuses. Post enumeration checks tend to be subject to many of the same limitations as the initial census. It would be useful to utilise the field organisation of the National Sample Survey to undertake a truly independent check. A more or less permanent working group set up by a body like the Indian Council of Social Science Research would make possible a continuing dialogue between the census organisation and the social scientists and other users of census data. A research endeavour with high standards of scientific objectivity would help in the evaluation of census data which form the very foundation of the national statistical system.