OBJECTIVES: To audit the proportion of clinical trials that had been publically registered and, of the completed trials, the proportion published. SETTING: 2 major research institutions supported by the National Institute of Health Research (NIHR). PRIMARY AND SECONDARY OUTCOME MEASURES: The proportion of trials reporting results within 12 months, 24 months and 'ever'. Factors associated with non-publication were analysed using logistic regression. INCLUSION CRITERIA: Phases 2-4 clinical trials identified from internal documents and publication lists. RESULTS: In total, 286 trials were identified. We could not find registration for 4 (1.4%) of these, all of which were completed and published. Of the trials with a registered completion date pre-January 2015, just over half (56%) were published, and half of these were published within 12 months (36/147, 25%). For some trials, information on the public registers was found to be out-of-date and/or inaccurate. No clinical trial characteristics were found to be significantly associated with non-publication. We have produced resources to facilitate similar audits elsewhere. CONCLUSIONS: It was feasible to conduct an internal audit of registration and publication in 2 major research institutions. Performance was similar to, or better than, comparable cohorts of trials sampled from registries. The major resource input required was manually seeking information: if all registry entries were maintained, then almost the entire process of audit could be automated--and routinely updated--for all research centres and funders.