This study aims to compare the effects of two food voucher programs on household welfare in Indonesia, namely the BPNT and Sembako Program, focusing on Java Island. It uses household food expenditure share (FES) and household dietary diversity score (HDDS) as outcome variables representing household welfare. This study utilizes the 2017–2021 cross-sectional data from the National Social Economic Survey (SUSENAS) published by the Central Bureau of Statistics. The Propensity Score Matching (PSM) and Ordinary Least Squares (OLS) methods were employed. The main estimate employs a multi-arm treatment approach to compare both programs to a common control group. By conducting PSM, this study controls about 76 observed characteristics of households. In addition, this method introduces time-variant and time-invariant fixed effects in the OLS estimation. The results show that the BPNT and Sembako Program decrease household FES and increase HDDS. The food voucher programs have different effects across provinces in Java, with food voucher programs tending to have greater effects in provinces with lower initial levels of welfare. Notably, the higher flexibility of the food voucher in the Sembako Program yields a greater effect on households. The higher flexibility of the food voucher allows beneficiaries to choose their food items other than rice and eggs. Thus, in addition to being effective in reducing FES, the Sembako Program is more effective in increasing HDDS. The policy implications directed at the government can improve targeting accuracy, voucher value, and infrastructure for accessing food vouchers, in alignment with the objectives set for the program.