It has been common to anchor poverty lines in nutritional requirements, generally calorie norms. There are many ways in which these calculations are conducted, but broadly people in households are designated as poor if they are estimated not to command sufficient calories in their diet to meet their computed requirements. Various calorie based methods of computing poverty lines or poverty are calculated for India and Bangladesh for recent years, including Food Energy Intake and Cost of Basic Needs methods. These two methods are shown to be essentially similar once constraints are introduced on the cost per calorie of the FEI method. However, poverty aggregates based on these calorie anchored poverty lines do not show the same spatial or temporal patterns of ill-being as other indicators which are plausibly related closely with poverty. This paper argues that anchoring PLs in calories is not a reliable way to compute comparable poverty lines for different domains of social groups, geographical spaces, or time periods; that is they do not produce poverty lines which represent the same standard of living in these different domains, and consequently poverty comparisons based on these lines may be comparing not differences in poverty but differences in the standards by which poverty is assessed.