Samples of 16 key fish species targeted by oceanic prawn trawlers and oceanic and estuarine beach seiners in southeastern Australia were measured for up to 10 morphological dimensions that included: total (Wt) and eviscerated weights (EWt); total (TL), fork (FL), natural (NL), standard (SL), dorsal standard (DSL), and anal standard lengths (ASL); and maximum body height (MH), width (MW), and girth (MG). All partial lengths were indexed against TL to provide linear relationships and, where there were sufficient data, tested for intra‐specific sexual dimorphism. Nearly all species that were sexed demonstrated at least some morphological divergence, and especially for transverse measures; probably owing to different reproductive conditions. Sex‐specific or combined regressions of DSL and ASL and transverse measures versus TL were used to identify the most appropriate: (1) partial length for compliance purposes; and (2) mesh sizes and configurations for the different fishing gears used. Using relevant data describing the smallest commercial‐sized fish species (stout, Sillago robusta and red spot whiting, S. flindersii— 150mm TL) retained in prawn trawls, a new square‐mesh codend (made from 35‐mm mesh hung on the bar) was designed and constructed. The selectivities of this square‐mesh codend and a conventional diamond‐mesh design (40‐mm mesh) were then modelled and compared in a case study. The conventional codend was demonstrated to be non‐selective for the sizes of eastern king prawns and red spot whiting targeted. In contrast, the square‐mesh codend selected red spot whiting across narrow selection ranges (18.88–15.67 mm) and at lengths of 50% retention (143.39–146.32mm TL) that closely corresponded to the estimated maximum mesh opening. The square‐mesh codend also improved selection for eastern king prawns, without impacting on commercial catches. Based on these results, similar predictions were made for appropriate mesh sizes and configurations in oceanic and estuarine beach seines; both of which were demonstrated to comprise minimum mesh sizes that would be non‐selective for the sizes of all targeted fish.