Estimates of fish abundance from electrofishing surveys depend on accurate estimation of capture probability. We examine in this paper how estimates of capture probability and abundance of Atlantic salmon from multi-pass removal sampling can be improved by comparing the results of an experimental programme of closed electrofishing sites within selected rivers in west-central Norway, and those obtained from open electrofishing sites established for monitoring long-term juvenile Atlantic salmon population abundance within the Burrishoole catchment, western Ireland. We first establish that the Carle & Strub method provides a more robust estimate of population abundance than the Zippin and Seber methods. We then show how prior information on capture probability may be used to improve the accuracy of the abundance estimate in open sites. We also show that the use of prior information with single-pass electrofishing may improve the accuracy of the abundance estimate so that it is comparable with that of multi-pass electrofishing in terms of stock prediction while requiring less sampling effort