Background: About 2500 new systematic reviews are indexed annually on MEDLINE, of which about one-fifth are Cochrane reviews. Systematic reviews should be kept up to date to ensure their validity and relevance, but there is limited methodology to inform the review updating process [1]. Collaboration policy states reviews should be updated within two years, but a priority-setting approach may be more appropriate [2]. Objectives: To consider and implement methods for update prioritisation of systematic reviews. Methods: A collection of indicators/signals for the need to update a review were developed in consultation with recent literature [1,2]. Software was developed for predicting the likelihood each signal is triggered by the inclusion of new evidence. Prioritisation rankings were then produced for each review being considered. Systematic reviews by the Stroke Review Group and others were used to evaluate performance of the signals. Studies published within three years of the most recent study in a review were excluded and the software applied. Reviews were ranked based on individual and total number of signals triggered. These were compared with the actual findings of the reviews. Results: A total of 12 signals were evaluated on 48 reviews. The use of relative weights of new and old studies in an updated review was found to be a robust signal, applicable even in the presence of heterogeneity. When using a combination of signals, the ratio of weights and power based on change in magnitude of the effect or statistical significance should be considered. Conclusions: We have identified a promising signal for prioritising reviews in need of updating but further work is required to establish an optimal combination of signals and to refine methods to make evaluations without the need for initial literature searches. The authors hope that in the future reviews will incorporate a section that summarises the new evidence that would be required to overturn conclusions. A prospective pilot of the software is welcomed. References: [1] Moher D, et al. J Clin Epi 2007;60:1095-1104. [2] Shojania KG, et al. Ann Inter Med 2007;147:224-233.