The purpose of this study was to develop and test an analytical technique for describing variations in the antibiotic prescribing process for otitis media (OM) in ambulatory medical plans. DESIGN: Using medical, hospital, and pharmacy claims over an 18-month period, a computerized rule-based system identified which antibiotics were probable therapy for OM. For each patient and OM episode of treatment, the rulebased system recorded the ordinal sequencing of specific antibiotics, including first-line and fOllow-up choices, to treat the OM condition. SETTING: Three medical plans in the Midwest were studied: a Medicaid-funded IPA-model HMO; an employer-funded, indemnity-based PPO; and an employer-funded "at-risk" network. RESULTS: The number of OM antibiotics dispensed per subject was similar among plans. For first-line therapy, inexpensive penicillinase-sensitive penicillins, trimethoprim-sulfamethoxazole, and erythromycin-sulfisoxazole, ....