Consumer preferences for food and non-food items in Pakistan are frequently estimated by using data from the household surveys. However, structural change in consumer preferences, caused by changes in tastes, can be studied by using the annual time series data, a time series of cross sections, or the panel data. This paper uses Pakistan’s annual time series disappearance data for eight food commodities from 1972 to 1991 to study consumer behaviour. The existence and the nature of structural change is tested by using both the generalised axiom of revealed preference (GARP) and the first-difference LA/AIDS model. It turns out that GARP tests are low-powered as tests of structural change on our data-set. However, the results from the LA/AIDS model show a shift in consumer demand from gram (split) to chicken after 1982. The data set satisfies symmetry and homogeneity. The estimates of price and income elasticities are also consistent with economic theory. The implications of these results for policy are also discussed.