Size selection characteristics of five cod-ends (114-mm, 127-mm and 140-mm diamond, and 114-mm and 127-mm square meshes) were tested through alternate-haul experiments performed under commercial fishing conditions in 1988–1990, with a 76-mm diamond mesh cod-end as standard. Overall selectivity curves for five flatfish species (arrowtooth flounder, Atheresthes stomias, and Dover, Microstomus pacificus, english, Parophrys vetulus, rex, Glyptocephalus zachirus, and petrale sole, Eopsetta jordanii) and five rockfish species (shortspine thornyhead, Sebastolobus alascanus, Pacific Ocean perch, Sebastes alutus, and widow, S. entomelas, yellowtail, S. flavidus, and canary rockfish, S. pinniger) were estimated using the Share Each Length Catch Total (SELECT) method and four types of selection ogives (logistic, probit, negative extreme value and Gompertz) on weighted pooled length distributions. The negative extreme value curve provided the best fits to more than half of the data, implying asymmetry in the retention of small and very large fish. Lengths of 50% retention increased with increasing mesh sizes for both the diamond and square mesh cod-ends, with rockfish providing on average larger values than flatfish. In general, the 114-mm square mesh cod-end retained more small fish than the other cod-ends. Thus, compared to the currently used 114-mm diamond mesh, it is not a good alternative to reduce discard or retention of immature fish. On the other hand, the 127-mm square and diamond, and 140-mm diamond mesh cod-ends reduced the retention of immature fish for many of the species. Thus, these cod-ends seem acceptable for biologically conservative management regulations.