Most poverty calculations use income-based or consumption-based measures. We introduce a new, third measure of poverty status which is based on household expenditure. Expenditure is the theoretically correct measure of resources actually transferred into a given period in the life cycle. But the presence of illiquid service flows from durable goods, which can only be used to purchase that good in the minimum standard of living bundle and no other good, requires an adjustment both to resources and to the poverty threshold to allow a proper comparison of liquid resources to liquidity-adjusted thresholds. We also show that the expenditure impact of transfers on the poverty rate requires an adjustment for effects of transfers on precautionary saving and for their role as consumption insurance. Using the 2009-2022 Consumer Expenditure Survey, our liquidity-adjusted poverty measure yields a 2022 poverty rate of 8.3 percent, declining by 11 percent since 2009. Our measure trends in broadly similar ways to income- and consumption-based measures but only because differences in how thresholds and resources are calculated are offsetting. We also find that accounting for precautionary saving and the insurance role of transfers could make the anti-poverty impact of transfers somewhat smaller than that estimated ignoring these effects.