Fishery management plans for U.S. fisheries are required to specify status determination criteria (e.g. whether the stock is overfished and whether overfishing is occurring) and typically use harvest control rules to adjust target and limit fishing mortality and catch levels to prevent overfishing, achieve optimum yield and rebuild overfished stocks. The status determination criteria are based on the concept of the fishing mortality rate ( F MSY) that maximizes long-term catch as the upper limit on the allowable rate of fishing and the associated B MSY, the spawning biomass which produces MSY, is the target for rebuilding of overfished stocks. In practice, proxies for the biological reference points F MSY and B MSY are often employed. Although several methods exist for estimating these quantities, it is unclear which performs best. Simulation is therefore used to evaluate alternative estimators for these quantities. These estimators differ in terms of whether a stock–recruitment relationship is estimated, and whether a prior based on Bayesian meta-analysis is used as a penalty on steepness, a critical parameter of the stock–recruitment relationship. The simulations consider three life histories: a long-lived unproductive rockfish, a moderately long-lived and productive flatfish, and a hake, which is also moderately long-lived and productive, but exhibits highly variable recruitment. Results indicate that estimator performance varies among reference points. However, estimators of B 0, the average spawning biomass in the absence of exploitation, and stock depletion based on a fitted stock–recruitment relationship generally perform best. B 0 is estimated either better (the rockfish and flatfish) or similarly (the hake) to stock depletion. Estimating B MSY from the fit of the stock–recruitment relationship performed best for the rockfish and flatfish life histories; average recruitment estimators proved to be best for the flatfish and hake life histories. Proxy methods of calculating B MSY generally performed relatively poor in comparison to the non-proxy measures. The performance of estimators of biological reference points was generally better for the rockfish and flatfish life histories, which were similar, than for the hake life history. Estimator performance was generally poorer in the presence of high recruitment variability.