Visual surveys provide for non-invasive sampling of organisms and habitats in the marine environment. They are particularly important in deep-sea habitats which are vulnerable to damage caused by alternative destructive sampling devices such as bottom trawls. However, traditional visual survey platforms tend to have limited area coverage which is insufficient for mapping the vast expanses of the deep-sea, particularly for relatively sparsely distributed organisms such as fish. This thesis presents the development of a visual survey method capable of surveying large areas of the seabed in deep waters (> 200 m) using a towed video camera system. The area of seabed sampled was similar to that sampled by a bottom trawl, making the system effective for surveying fish. Anglerfish densities were found to be comparable to those determined by trawl surveys in adjacent areas. For other deep-sea fish, the two survey methods (towed camera and bottom trawl) provided different results which were explained by the behavioural reactions of different fish taxa to the respective gears. Fish that exhibited detectable avoidance behaviour to the towed video camera system (e.g. Chimaeridae) resulted in significantly lower density estimates than trawl estimates. On the other hand, skates and rays showed no reaction to the towed video camera system, and density estimates of these were an order of magnitude higher than the trawl. The towed video camera survey was also deployed to gather data on the cold-water coral Lophelia pertusa on the Rockall Bank. These data were used together with Species Distribution Modelling (SDM) to predict the distribution of Lophelia pertusa habitat. The current closed areas on the Rockall Bank cover sizeable areas of potential Lophelia pertusa habitat, however, based on the predictions further areas could be considered to ensure the continued protection of this species.