Before a multicentre trial of 3-D conformal radiotherapy to treat cancer of the pancreas, participating clinicians were asked to complete an accreditation exercise. This involved planning two test cases according to the study protocol, then returning hard copies of the plans and dosimetric data for review. Any radiation technique that achieved the specified constraints was allowed. Eighteen treatment plans were assessed. Seven plans were prescribed incorrect doses and two of the planning target volumes did not comply with protocol guidelines. All plans met predefined normal tissue dose constraints. The identified errors were attributable to unforeseen ambiguities in protocol documentation. They were addressed by feedback and corresponding amendments to protocol documentation. Summary radiobiological measures including total weighted normal tissue equivalent uniform dose varied significantly between centres. This accreditation exercise successfully identified significant potential sources of protocol violations, which were then easily corrected. We believe that this process should be applied to all clinical trials involving radiotherapy. Due to the limitations of data analysis with hard-copy information only, it is recommended that complete planning datasets from treatment-planning systems be collected through a digital submission process.