Abstract: There has been a good deal of difference of opinion about the actual level of con sumption of cereals in India, especially since the National Sample Survey1 (NSS) started making estimates of per capita consumption about ten years ago. The estimates of aggregate consumption were obtained by multiplying the total population by the per capita consump tion; and such estimates were found to be consistently much higher than the estimates of production of the Ministry of Food and Agriculture which were based essentially on the so-called 'complete enumeration' of the area sown by patwaris (employees of the Revenue Department) at the village level. The NSS estimates based on per capita consumption have been consistently higher by something of the order of twenty-five per cent or more. It is pertinent to mention that traditional estimates, based on complete enumeration, when checked by sample surveys or compared with other independent estimates, have shown similar underestimation even in the U.S.A.; and a similar gap in Japan was reported in a paper presented last year (1960) at the Tokyo Session of this Institute.2 2. Aggregate estimates of consumption based on per capita consumption involve the use of the estimate of total population. In India the estimate of population based on the Census of this year (1961) is now available; it is, therefore, possible to calculate the aggre gate consumption based on the Census figure. Very recently, the National Sample Survey has obtained independent estimates of production of cereals for India as a whole based on sample surveys, by direct physical observation of the area sown in sample plots of land, and estimates of yield per acre based on direct crop-cutting experiments using probability samples. Also, as a design of 'inter-penetrating network of samples' (IPNS) is used in the NSS, at least two independent estimates are available; and the difference between the two estimates supplies the margin of uncertainty which can be used for statistical comparisons in a valid manner.