Background: A small number of prescriptions ordered by the patient from their general practice remain uncollected and hence undispensed. No research has been published on this phenomenon and on how this is managed. We aimed to evaluate this in one primary care trust. Aim: To find out what prescription items are not collected, and why. Design of study: A descriptive cross-sectional analysis of prescription data. Semistructured interviews with 21 primary health care team members, and 10 patients who had apparently not collected their prescription. Fifty-seven patients from the lead author's practice were telephoned and gave their comments. Setting: Twenty general practices in the Gateshead Primary Care Trust. Method: Nineteen practices provided suitable data for analysis from one month's uncollected prescriptions plus total items issued during the same period of time. All suitable patients who had uncollected prescriptions from 10 practices were invited to participate in a telephone interview. Similar patients from the lead author's practice were telephoned and invited to comment. Results: On average 0.5% items were uncollected. Drugs for a specific diagnosis (eg, cardiovascular drugs) were significantly less likely to be uncollected than drugs prescribed either symptomatically or for a presumptive diagnosis (0.48% versus 0.67% uncollected, respectively, P , 0.001). Many uncollected prescriptions were due to administrative causes: few resulted from patient error or forgetfulness. The majority of patients reported obtaining their medication. No adverse events arising from uncollected prescriptions were reported. Conclusion: Uncollected prescriptions are a small proportion (0.5%) of the total issued and were more likely to be for non-essential items; therefore the policy of destroying uncollected prescriptions after an appropriate period without any further action is probably safe.