The wide range of calorie-income elasticities in the literature results, in large part, from the particular calorie and income variables used for estimation. Elasticities across four estimation techniques and four calorie-income variable pairs for a sample of Philippine farm households, ranged from 0.03 to 0.59. Estimates associated with calorie availability are biased upwards, first, because random errors in measuring food purchases are transmitted (by construction) both to calorie availability and total expenditures, and second, because the residual difference between family calorie intake and household calorie availability will often increase with income. The calorie intake-total expenditure variable pair gives the preferred elasticity estimate in the 0.08-0.14 range.