BACKGROUND: Despite evidence that uncomplicated lower respiratory tract infection (cough) does not respond appreciably to antibiotics and that bacterial resistance is increasing, general practitioners (GPs) still prescribe frequently. AIM: To assess delayed antibiotic prescribing as a strategy for reducing the unnecessary use of antibiotics for cough in primary care. DESIGN OF STUDY: Open randomised controlled trial of delayed versus immediate prescribing of antibiotics. SETTING: One hundred and ninety-one adult patients with uncomplicated cough in 22 Scottish practices who would have received antibiotics under the GP's usual practice were randomised to receive either an immediate prescription (92 patients) or a delayed prescription (99 patients). METHOD: Delayed subjects were asked to wait a week before deciding whether to collect their prescription. Outcome measures included symptom duration, prescription uptake, patient satisfaction, patient enablement, and subsequent consultation rates. The 48 GPs who recruited patients were surveyed six months after the trial to see whether they used delayed prescribing as a part of their normal practice. RESULTS: Study and control groups were similar at baseline. Of the subjects in the delayed arm, 55% did not pick up their prescription. Although most patients were satisfied, more patients in the immediate arm were very satisfied with the treatment (P = 0.001) and the consultation (P = 0.03). The patients in the immediate arm were also more enabled (3.3 versus 2.4; P = 0.04), although more of them intended to consult for similar complaints in the future (85% versus 69%, P = 0.02). We were unable to detect any difference in actual consulting behaviour in the follow-up period (mean = 15 months [SD = 5 months]). Subsequently, 68% of GPs used delayed prescribing at least monthly; all gave the prescription to the patient. CONCLUSION: Delayed prescribing is effective at reducing the use of antibiotics for self-limiting cough; however, patients are less satisfied and enabled as a result. Patients may be deterred from consulting rather than becoming enabled.