In this paper we estimate a model of household labor supply using an econometric approach that allows simultaneous decisions of household members, complex and non-convex choice sets induced by tax and benefit rules and quantity constraints on hours choice. The model is estimated using the 1993 Bank of Italy's survey of household income and wealth and used to simulate three hypothetical tax reforms, namely, a flat tax and two versions of a negative income tax system, under the constraint of equal tax revenue. All these reforms contain both incentives to work less and incentive to work more. The incentives to work more seem to prevail, at least for the more productive: the labor supply elasticities, although modest, are sufficiently large (especially female participation elasticity), so that all the reforms produce a larger household average disposable income, without worsening much the equality of the income distribution as measured by the Gini index. A remarkable result is that no significant `poverty trap' effect is associated with the negative income tax reforms. All the reforms are supported by a majority of winners in the sample, although the proportion of winners varies considerably across income deciles.