In international achievement studies, questionnaires typically ask about the presence of particular household assets in students' homes. Responses to the assets questions are used to compute a total score, which is intended to represent household wealth in models of test performance. This study uses item analysis and confirmatory factor analysis to assess the validity of assets indices as wealth measures in three educational surveys of primary or secondary school students. We find that scores generated from binary household asset items can reasonably be interpreted as predominantly measuring a single wealth factor, but some scores are imprecise, with between one-third and one-half of their variability attributable to measurement error. Our study illustrates the potential of psychometric analyses to contribute to improvement of the household wealth items administered in similar international achievement assessments.