ObjectiveTo determine patterns of use of cord blood gas analysis in two institutions in Portland, Maine; to determine which factors, if any, predicted use of the test; and to evaluate compliance with ACOG guidelines.MethodsBilling data were used from 3166 deliveries during 1994 in the two hospitals to find deliveries in which the test was performed. We merged billing data with birth certificate data to examine factors associated with the test's use. Finally, we compared its use in our community with recently updated COG guidelines.ResultsThere was a 20-fold difference in the test's use between institutions (P< .001). The test was performed in 49% of all births at the tertiary care center and 2.5% of births at the community hospital. Many maternal and neonatal factors were linked to use of the test, but delivery system factors, in particular, the institution, were the strongest predictors of the test's use, even controlling for confounding factors between hospitals. There was little adherence to ACOG guidelines with 97% of the tests being performed in situations in which ACOG's recommendations did not support use of the analysis.ConclusionCurrent use of cord blood gas analysis in our community is not consistent with guidelines or cost-effective use of resources; the institutional factors that determine excess use should be examined and modified.