Fishing effort in “traditional” creel surveys is derived from instantaneous angler counts, which in some fisheries can carry a high cost in personnel time, resources, and safety. To estimate angler effort more efficiently and at lower cost, we used mark–recapture methods to estimate weekly angler populations in Pacific salmon Oncorhynchus spp. and steelhead O. mykiss fisheries on the Salmon River, Idaho. Weekly harvest estimates were the product of weekly angler population, mean number of days fished per angler per week, and mean harvest per angler per day. We compared traditional and mark–recapture estimates of weekly harvest using paired analysis on a sample of 48 weeks. We found no significant difference in harvest estimates between traditional and mark–recapture methods, and the mark–recapture estimates in our fisheries could be made at a savings of up to 50% in vehicle mileage associated with conducting angler counts. However, the width of 95% CIs around harvest estimates was significantly higher for the mark–recapture method, due to higher upper confidence limits resulting from right‐skewness of the sampling distribution of the mark–recapture‐based estimator. Precision could be improved by replacing time spent on counts in the traditional method with more time spent “capturing” anglers at access sites, still providing savings in vehicle mileage.