“Calibration” is a weighting procedure performed by many household surveys that modifies the sample households’ weights to make them add up to certain known population values. In practice the number of known population values must be small to avoid small sample size problems like empty cells. When the number of known population values that are available exceeds the number that are usable, a decision needs to be made about which ones to use. Naturally, one wants to use the best set of known population values, but that requires knowing what the word “best” means. In this paper we review what various authors consider the goals of the calibration procedure to be, and how these goals apply to the Consumer Expenditure Survey.