This paper explores the statistical properties of household consumption-expenditure budget shares distributions (HBSDs) defined as the share of household total expenditure spent for purchasing a specific category of commodities for a large sample of Italian households in the period 1989-2004. We find that HBSDs are fairly stable over time for each specific category, but profoundly heterogeneous across commodity categories. We then derive a parametric density that is able to satisfactorily characterize HBSDs and: (i) is consistent with the observed statistical properties of the underlying levels of household consumption-expenditure distributions; (ii) can accommodate the observed acrosscategory heterogeneity in HBSDs. Finally, we taxonomize commodity categories according to the estimated parameters of the proposed density. We show that the resulting classification is consistent with the traditional economic scheme that labels commodities as necessary, luxury or inferior. -- Household Consumption Expenditure ; Budget Shares ; Sum of Log-Normal Distributions