Effective means of affecting physician prescribing behavior must be developed if the quality of health care is to be improved. A drug therapy protocol for otitis media, containing a rationale for drugs, doses and days of therapy by patient age, was developed by a physician panel at one site (EI) of an HMO, and implemented at both (EI); and a second site (EII); a third site was a control. A comparison of prescribing during two phases, before and after implementation of the protocol, suggested that the protocol had a greater influence at site (EI); than site EII, and that it had a greater influence on the panel physicians than on other prescribers at site EI. Questionnaire responses indicated that clinical experience and personal involvement were the most important protocol use influences. If participation is critical to protocol compliance, the protocol method to improve the quality of health care is severely limited.