Discarding is the practice of returning unwanted catch to the sea during commercial fishing. However, the discarding process is costly and time consuming and some fish bycatch species have a high post-capture mortality rate. Therefore, reducing bycatch and thus discarding is a major reason for ongoing research and development into Bycatch Reduction Devices (BRDs). This research presents a novel BRD that uses artificial light attached to the headline of commercial prawn trawl nets and insight into its effect on reducing the overall capture of fish bycatch. Two designs of light BRDs were developed and tested in the temperate waters of North-Eastern Tasmania as well as the tropical waters of Moreton Bay, Queensland. It was found that there was a 50% reduction in total fish bycatch in temperate waters but no evidence of a significant difference in total fish catch in tropical waters. There were species-specific differences in the changes of catch rates with the use of the lights. Most species were found to decrease in catch with the use of light, and one species in particular, P. bassensis, was reduced by 75%. The only evidence that the lights had any effect on size distribution was found for two species, T. declivis and P. melbournensis that had significantly different length frequency distributions with the use of artificial light. In an attempt to explain species-specific changes in catch rates, the visual and swimming capabilities of a range of bycatch species were studied. The photoreceptor cell densities and potential visual acuity were quantified using histological techniques on the eyes of eight species of interest: Lepidotrigla mulhalli; Lophonectes gallus; Platycephalus bassensis; Sillago flindersi; Neoplatycephalus richardsoni; Thamnaconus degeni; Parequula melbournensis; and Trachurus declivis. The cone densities ranged from 38 cones per 0.01 mm2 for S. flindersi to 235 cones per 0.01 mm2 for P. melbournensis. The rod densities ranged from 22 800 cells per 0.01 mm2 for L. mulhalli to 76 634 cells per 0.01 mm2 for T. declivis and potential visual acuity (based on anatomical measures) ranged from 0.08 in L. gallus to 0.31 in P. melbournensis. Higher rod densities were correlated with maximum habitat depths. Parequula melbournensis had the greatest potential ability for detecting fine detail based on eye anatomy. The stride length and maximum swimming speeds were estimated for five of these eight species. The maximum swimming speeds of L. mulhalli, P. melbournensis, P. bassensis, T. degeni and T. declivis were 1.71, 4.17, 4.80, 3.19 and 6.40 m s-1, respectively. Trachurus declivis had the longest stride length and fastest maximum swimming speed. Therefore, based on swimming capability it is predicted that of the five species studied, T. declivis would be most likely to avoid capture by a trawl net. The results show a linear relationship between the potential visual acuity and percent change in catch rate, and also between the maximum swimming speed and percent change in catch rate of the species of interest. Maximum swimming speed explained 83 % and 88 % of the change in weight and numbers, respectively. Potential visual acuity was only able to explain 5 % and 23 % of the change in weight and numbers, respectively. When combining the two factors, they accounted for 74 % and 82 % of the change in weight and number, respectively. This study concludes that maximum swimming speed is sufficient for predicting the percent change in catch rate of a species when using artificial light. This relationship is beneficial for predicting the catch rates of different species in trawls fitted with the novel light BRD.