Accuracy of angler-reported data on steelhead, Oncorhynchus mykiss (Walbaum), harvest in Idaho, USA, was quantified by comparing data recorded on angler harvest permits to the numbers that the same group of anglers reported in an off-site survey. Anglers could respond to the off-site survey using mail or Internet; if they did not respond using these methods, they were called on the telephone. A majority of anglers responded through the mail, and the probability of responding by Internet decreased with increasing age of the respondent. The actual number of steelhead harvested did not appear to influence the response type. Anglers in the autumn 2012 survey overreported harvest by 24%, whereas anglers in the spring 2013 survey under-reported steelhead harvest by 16%. The direction of reporting bias may have been a function of actual harvest, where anglers harvested on average 2.6 times more fish during the spring fishery than the autumn. Reporting bias that is a function of actual harvest can have substantial management and conservation implications because the fishery will be perceived to be performing better at lower harvest rates and worse when harvest rates are higher. Thus, these findings warrant consideration when designing surveys and evaluating management actions.