The measurement of per capita daily expenditures which are compared with a monetary poverty line is the most widely used approach regarding poverty assessment. It is, however, based on the implementation of time and cost-intensive household surveys and, therefore, not an operational method for targeting poor households with development services. The paper shows how to identify an alternative poverty assessment tool for Peru. It consists of a maximum of 15 powerful predictors of per-capita household expenditures selected out of a wide range of indicators from different poverty dimensions such as education, assets and housing characteristics. By applying the maximizing-R-squared regression technique to identify the best 5 to 15 predictors, we avoid an arbitrary indicator selection and the application of external weights. In a second step, an innovative approach based on the percent point function of the predicted expenditures is used for the poverty classification of households. The resulting poverty classification of households, as validated by different accuracy measures and their 95% confidence intervals, reveals that the 15 indicator tool correctly identifies over 81% of the poor households when taking the national poverty line as the relevant benchmark. The high accuracy in terms of its predictive power is confirmed by out-of-sample tests and suggests that the tool is an interesting alternative to the collection of detailed expenditure data. Before employing the tool in practice, the indicators still have to be tested for their robustness across time and then transformed to a short, focused questionnaire suitable for both ex-ante poverty targeting and ex-post impact assessment.