Although complex pricing schedules are increasingly common in utility billing, it is difficult to determine whether consumers respond to complicated marginal prices because price changes are often confounded with simultaneous demand shocks or non-price policies. To overcome this challenge, we exploit a natural experiment - the introduction of a third price block in an increasing block pricing schedule for water - in Santa Cruz, California. Using a regression discontinuity design, we find that consumers do respond to changes in marginal price. Doubling marginal price led to a 12% decrease in water use (500 cubic feet per bill) among high-use households.