We used simulation modeling to explore how three statistical catch-at-age approaches for assessing intermixed fisheries performed in terms of assessment accuracy and management performance, under differing productivity, mixing, and harvest levels. Simulations were based on intermixing lake whitefish (Coregonus clupeaformis) populations in the upper Laurentian Great Lakes of North America. We found that with intermixing, the “separate” assessment approach, which ignored intermixing and treated mixed populations as unit stocks, produced biased estimates of spawning stock biomass (SSB); however, the “pooled” assessment approach, which lumped populations and assessed them as a single stock, was nearly unbiased in estimating SSB. The “overlap” assessment approach, which estimated the populations in one combined assessment model by incorporating actual mixing rates, was most strongly biased in estimating SSB in the absence of mixing, with bias decreasing as mixing levels increased. With high mixing levels, the overlap method had difficulty converging on unique solutions. The pooled approach provided better management performance than the separate approach with intermixing. When the overlap method could be applied, it provided the greatest SSB with little reductions in yield and the lowest inter-annual variation in yield. Relative performances of the assessment approaches were robust to assumed harvest levels.