We wanted to assess whether routine use of a rapid test for C-reactive protein (CRP) could reduce prescription of antibiotics for adults with possible lower respiratory tract infection. 239 patients were randomized into a CRP group, tested with the rapid test (n = 108) and a control group (n = 121). Before knowing to which group the patient belonged the doctors made a preliminary decision about antibacterial treatment. The C-reactive protein value was then released if the patient belonged to the CRP group, and the therapy could be adjusted in light of the result. Antibacterial courses prescribed during the consultations and in the following three weeks were registered. The clinical course was evaluated by interview after one week and again after three weeks. Antibiotics were prescribed for altogether 56% of the patients in the CRP group and 60% in the control group. The difference was not statistically significant. Prescription of antibiotics was strongly associated with the finding of crackles and wheezes, but not with cough, dyspnoea or chest pain. Slow recovery was associated with high age, absence of fever and a normal value of C-reactive protein. No significant benefit of the CRP test was demonstrated. We discuss whether the doctors made full practical use of the information provided by the test. Bronchial obstruction should probably be considered to be the problem more often in coughing patients with a normal CRP value.