The Acadian redfish Sebastes fasciatus stock is one of the few healthy fished populations in the Gulf of Maine. This stock recovered from decades of unsustainable exploitation in the mid-20th century after imposition of codend mesh size restrictions and a long period as a bycatch fishery. Current harvest levels are far below permitted levels, and therefore the stock potentially offers additional opportunity for regional fishermen. However, based on its history and unusual biology, re-establishment or expansion of the trawl fishery for redfish should be conducted with caution. To assure future health of the stock, steps to encourage additional landings should include consideration of a wide range of aspects, including appropriate fishing gears, timing of seasons, processing, and marketing, as well as population dynamics. This dissertation sought to provide several vital conservation engineering or fishing technology inputs to a sustainable fishery with optimum yields in a stepwise fashion, beginning with identification of bycatch issues, followed by codend selectivity, and bycatch reduction. This approach, combined with data from other sources, is adequate to inform careful decision making about increased exploitation of the stock. As a first step, basic catch and bycatch data were collected by commericial fishing vessels, every two to three months between May 2011 and January 2012. The vessels used their own commercial groundfish trawls, but a relatively small mesh codend (114 mm or 4.5 in mesh size); catch and bycatch were documented by onboard observers. The resulting catch consisted predominately of redfish (105,627 kg; 84.6% of all catch). Pollock (Pollachius virens) was the main landed bycatch species (3,322 kg), with 13 other species landed in smaller amounts. Discard rates using a modification of the Standard Bycatch Reporting Method revealed that almost all were below 0.01 kg if bycatch per kg of kept catch. These results indicate harvest of redfish using 114 mm diamond mesh can occur without substantial catch of undersized redfish and other commercially important groundfish species, although rates for very weak stocks, such as Atlanic cod Gadus morhua could be problematic. Depth, time-of-year, or some other factors appeared to have an impact on the catch of undersized redfish but could not be isolated. Next, codend size selectivity for redfish and pollock were investigated using an innovative trouser trawl section to determine retention curves for three sizes of mesh opening (114, 139 and 165 mm double 5 mm twine diamond) on a commercial fishing vessel. Fifty-six tows were completed in March and April 2013. Robust selectivity models for the mean L₅₀s and selection ranges (SR), and confidence intervals, were developed for all three tested codends, incorporating both within and between haul variability: nominal 114 mm - L₅₀: 22.3 cm (redfish) and 34.8 cm (pollock); 139 mm - L₅₀: 29.2 cm (redfish) 45.6 cm (pollock); 165 mm - L₅₀: 33.6 cm (redfish) and 52.4 cm (pollock). A simple relationship between the L₅₀s for the two species was established. These models can guide managers and fisherman on mesh size and retained sizes of redfish and its primary bycatch. Additionally, simulation of fishing of the three tested codends on the observed population indicated that substantial escape of redfish through codend meshes occurs (51-96% of all redfish that enter the trawl). The observed size structure of the redfish in combined catches also indicates that inadequate numbers of larger redfish may be available to support a higher-priced market. The final field phase investigated timing and location of codend escapes by redfish using cameras and lights, and a dual grid system to increase escape of small redfish at depth. Escapes of redfish were observed on video to occur at all depths, but most frequently during net retrieval, exposing smaller redfish to more potential sources of mortality. Cameras recorded as many as 50% of estimated escapes. Testing of two bar spacings for a dual grid system with a 114 mm codend demonstrated the possibility of improving size selection through the use of size-sorting grids. While the 40 m grid showed no differences in size structure of the redfish catch, the 50 mm spacing resulted in reductions all sizes of redfish in the catch. Population analysis modeling possible effects of depth-differentiated estimated rates of escape mortality showed strong impacts of fish escapes on population trends in the stock; these impacts were sensitive to the estimated rates, which were based on redfish physiology and other factors. Additional modeling examining the impact of changing the fishery selectivity indicated that the largest mesh size of 165 mm maximized yield per recruit. All fishing related results from this dissertation were combined with other sources to illustrate tradeoffs in management decisions. A primary result is that, under the status quo, which includes options to use 139 or 165 mm codends, exploitation levels will likely not reach the target level. Also, use of larger meshes increases numbers of redfish escaping the codend and likely increases unobserved mortality, and can adversely affect the population size estimates. Recommendations include support for a monitored special access program with broad geographic and temporal scales, with a 114 mm codend permitted. In the longer term, refinement of a grid system, quantification of redfish codend escape mortality rates and reduction of bycatch of pollock and cod are recommended.