In most electricity systems the residential sector is one of the main contributors to the system peak. This makes it important to know how different residential end uses, such as space heating or cooking, contribute to the system load curve at the time of system peak and also at other times of the day. In this paper we discuss the estimation of residential end-use load curves for the state of New South Wales in Australia. Half-hourly readings were taken for 15 months on the total load and a range of end-use loads of 250 households. Information was sought on 16 different end uses, while eight metering channels were available for each household. We describe the optimal design procedure used to determine which end uses to meter in each household. The econometric model used for estimating the end-use load curves integrates a conditional demand analysis (CDA) of the total load readings for the household with the readings on all the directly metered end uses. Our integrated approach achieves impressive gains in efficiency over the conventional approach to estimating end-use loads. The paper concludes with an illustration of how end-use load curves can be used to simulate a variety of policy options.