Incorporating catch or harvest rate information in repeated-choice recreation fishing demand models is challenging, since multiple sources of information may be available and detail on how harvest rates change within a season is often lacking. This article develops a theoretically consistent harvest expectations repeated mixed logit angling demand model that can be used to evaluate the contributions made by different sources of information in predicting observed patterns of fishery participation and trip frequency. In an application to saltwater salmon fishing in Alaska, we find that both of the two available harvest rate information sources contribute to better predictions and should be used. In addition, information on whether a species is being targeted makes a significant improvement to model performance. Model tests indicate that non-targeted species have a significant marginal utility and it is different from the marginal utility of targeted species. The median value of a fishing choice occasion is approximately $49 per angler, which translates to a season of fishing being valued at a little over $2,400 at the median per angler.

JEL Classification Codes: Q51, Q22, Q58, C35