To test the validity of a nurse-administered protocol for low back pain, a prospective trial of 419 patients was undertaken in a walk-in clinic. In all, 222 patients were randomly allocated to a “nurse-protocol group” in which they were evaluated by one of five nurses using the protocol; the nurses independently managed 53 percent of the patients and referred to a physician patients with potentially complex conditions. In addition, 197 patients in a randomly allocated control group were managed by one of 32 physicians. Care in the experimental and control groups was compared by follow-up telephone contact and by a four-month chart review. There was no significant difference in symptomatic relief or the development of serious disease in the two groups. Nurse-protocol patients expressed greater satisfaction with the care they had received; patient satisfaction correlated positively with symptom relief. In over 95 percent of the patients, there were noncomplex, nonserious, nonchronic conditions as the cause of back pain. We conclude that nurse-protocol management of this generally benign condition in a primary care setting is both effective and efficient.