Selective exploitation can cause adverse ecological and evolutionary changes in wild populations and also affect sex ratios but few studies have empirically documented skewed sex ratios in exploited fishes (other than species with extreme sexual size dimorphism, SSD). To investigate the possibility of sex-selective fishing on Alaskan sockeye salmon Oncorhynchus nerka, we assessed sex ratios in fish at two spatial scales: within each of five fishing districts and among 13 breeding populations in one of these districts. We predicted that populations’ sex ratios would vary based on the average size of fish and SSD because size affects vulnerability to fishing. At the larger scale, we found a small but significant bias in fish returning to four of the five fishing districts (average = 52% females), and in four of the five districts males were caught at significantly higher rates than females. At the finer scale there was marked variation in sex ratio on the breeding grounds, ranging from 36% to 47% males. Populations with fish of intermediate sizes experienced the greatest sex ratio biases; the greater vulnerability of males than females to fishing resulted from a combination of larger SSD and different harvest rates between the sexes associated with the fishery size-selectivity curve shape. Skewed sex ratios may change competition and behavior on the breeding grounds, relaxing selection on male traits associated with mate choice by females or intra-sexual competition and altering demographic and evolutionary pressures on the fish. Assessment of the size selectivity of fishing gear and the population's SSD can help to illuminate if and how exploitation can affect sex ratios. Future studies examining size-selective fishing should also evaluate the consequences for sex ratios, as this might help explain changes in harvested population structure and sustainability.