Background: The Healthy Purchase Index (HPI) assesses the nutritional quality of food purchases (FP) from food group expenditure shares only. However, it was developed from the FP of a disadvantaged population. Objective: To adapt and validate the HPI for a general population. Methods: FP were obtained from a representative sample of French households (Kantar WorldPanel) subdivided into two subsamples. The first sample (n = 4375) was used to adapt and validate the score; the second sample (n = 2188) was used to test external validity. The revised-HPI (r-HPI) includes 2 subscores: the diversity subscore and the quality subscore. Diversity subscore points were awarded when expenditure shares were above the 25th percentile for 5 food groups ("Fruits", "Vegetables", "Starches", "Dairy", "Meat, Fish and Eggs"). Regression models between the expenditure shares of each food group and the Mean Adequacy Ratio (MAR) and the Mean Excess Ratio (MER) of FP were used to select quality subscore components and define cut-offs for point allocation. Construct validity was assessed on the first sample using Spearman's correlations between the r-HPI and the four nutritional quality indicators (NRF9.3, MAR, MER, energy density), and also by comparing the r-HPI of monthly FP of sub-populations defined by criteria known to influence diet quality (age, gender, income, education) and between households having a monthly food basket of higher (MAR > median and MER and energy density < median) vs. lower nutritional quality within the population, using Wilcoxon tests or pairwise comparisons of contrasts. External validity was tested by performing the same analyses on the 2nd sample of 2188 households. Results: The adaptation led to include new components (e.g. red meat) and define new cut-offs (e.g. - 1 point when budget share for red meat > 21%). The r-HPI (mean = 6.50 ± 3.58) was strongly correlated with NRF9.3, MAR, MER and energy density (0.59, 0.52, - 0.41 and - 0.65, respectively, p < 0.01) and poorly correlated with total energy content (- 0.096, p < 0.001). The r-HPI was significantly higher in women (β = 1.41 [0.20], p < 0.01), households having a food basket of higher nutritional quality (β = 4.15 [0.11], p < 0.001), and increased significantly with age, income and education levels. Similar results were obtained in the 2nd sample. Conclusion: We showed the validity of the r-HPI in a large sample of French households. As it does not require food quantity or nutrient content, it can be used as a valuable tool to explore FP behaviours. Cut-offs can be used in health promotion to provide nutri-economic counselling.