Objective To develop an index to assess the nutritional quality of household food purchases based on food expenditures only. Design A database of monthly food purchases of a convenience sample of low-income households was used to develop the Healthy Purchase Index (HPI). The HPI is the sum of two sub-scores based on expenditure shares of food categories in total household food expenditure: the purchase diversity sub-score and the purchase quality sub-score. The first was adapted from an existing diversity score. The second integrated those food categories identified as the best predictors of the nutritional quality of purchases based on associations between expenditure shares of food categories and two nutritional quality indicators: the mean adequacy ratio (MAR) and the mean excess ratio (MER). Correlation between the HPI and a score assessing adherence to French dietary guidelines (PNNS-GSmod) was performed as a first validation. Setting/Particpants Food purchases of 112 households from deprived neighbourhoods of Marseille (France), participating in the Opticourses and Jassur projects (2012–2015). Results The purchase diversity sub-score reflects the presence in food purchases of five food categories: fruits, vegetables, starches, dairy products, and meat, fish & eggs. The purchase quality sub-score is based on expenditure shares for fruit & vegetables, added fats & seasonings, sweet snacks, cheese, sugary drinks, refined grains and fish, as these were identified as predictors of the nutritional quality of purchases. The HPI was positively associated with the PNNS-GSmod (rs=0·378; P<0·001). Conclusions The HPI helps assess the healthiness of household food purchases.