This paper examines whether delivery practice patterns (Cesarean sections or vaginal) are influenced by lawsuits or whether the hypothesized relationship is confounded by small area variation. The analysis uses multilevel analysis to deal with hospital- and Dartmouth Hospital Referral Region-level variation. The model includes patient clinical variables, patient socio-economic status, and hospital characteristics as control variables. The secondary data sources include hospital discharges from the 2002 Texas Health Care Information Council as well as 1988-2001 Texas Department of Insurance Closed Claim File data. After extracting the variation in delivery practice between hospitals and between Dartmouth Hospital Referral Regions in a multilevel model, the effects of lawsuits on defensive medicine are reduced but are still significant.