We present the results of an experiment involving established couples, which uses choices between lotteries to test some economic models of household decision-making. Subjects make choices individually and jointly and are asked to make predictions about their partner's choices. Income pooling is not rejected in joint choice but has less explanatory power in individual choice. Many joint choices do not satisfy the Pareto principle. Overall, couples are more risk averse when making choices jointly compared to individual choice. Gender is not a direct determinant of power in joint choices, but female economic dependence significantly reduces women's decisiveness in joint choice.