The simultaneous operation of per case and per service payment systems in Maryland, and the varying levels of stringency used in setting per case rates, allows a comparison of the effects of differing incentive structures on hospital costs. This paper presents such a comparison with 1977-1981 data. Regressions performed on cost-per-case and total cost data indicate that costs were lower only when per case payment limits were very stringent. Positive net revenue incentives appeared to be insufficient to induce a reduction in length of stay or ancillary services use. These changes in medical practice patterns thus appear more likely under the threat of financial losses--that is, under the threat of the stick rather than the inducement of the carrot.