Demand-side response (DSR) measures, which facilitate the integration of high shares of intermittent renewable generation into electric grids, are gaining prominence. DSR measures, such as time-of-use (TOU) rates, charge higher rates during high-demand 'on-peak' times. These rates may disproportionately impact the energy bills and health of vulnerable households, defined as those who face greater energy needs combined with greater social and financial pressures. Here we examine 7,487 households that took part in a randomized control TOU pilot in the southwestern United States. We found that assignment to TOU rather than control disproportionately increases bills for households with elderly and disabled occupants, and predicts worse health outcomes for households with disabled and ethnic minority occupants than those for non-vulnerable counterparts. These results suggest that vulnerable groups should be considered separately in DSR rate design, and future pilots should seek to determine which designs most effectively avoid exacerbating existing energy injustices or creating new ones.