This paper investigates the conditions under which expected equivalent variation provides a correct ranking of stabilized and uncertain price situations. Expected equivalent variation can be recovered from ordinal preferences whereas ex ante equivalent variation requires additional information regarding con sumer attitudes toward income risk. The two measures are equal for income risk-neutral consumers. More important for applied welfare analysis where income risk aversion is presumed, expected equivalent variation provides a lower bound for ex ante equivalent variation. These two features of expected equivalent variation commend it as a viable welfare measure for applied work.