Banking as a financial institution and a business has gone beyond providing banking servicesand making profits for the proprietors to assume macro positions like involving in activitiesthat propel an entire economy. This has earned the interest of national governments and thesociety as well as the international community as a whole inciting a need for supervision inorder to ensure sustainability.Banks as corporate organizations with the above stipulated stake holders and management arein a constant tog of war with each stakeholder seeking to protect its own kind which hasdegenerated to a vice often known in corporate governance as agency cost or principal-agentconflict. In corporate governance there is a board of directors that is designed to align theshareholders interest and management interest in order to check this agency cost.In our thesis we have expatiated widely on the concept of corporate governance, board ofdirectors and its composition. We have isolated gender diversity, which is one of thecompositions of the board to find out how it contributes to control agency cost by establishingits effect on ROE, which is a firm performance indicator. We introduced control variables tocheck our results.We collected data from the annual reports of Nordea, Swedbank, Handelsbanken and SEB,which are the listed banks in the OMX NASDAQ exchange Stockholm for period 2001 to2010. The board characteristic we used is the proportion of females in the entire board and thefirm performance indicator we chose is ROE. We carried out a longitudinal study for thewhole industry and for individual firms in the industry over the ten-year period.We have variation in the results over the different firms and in the industry but there is nosignificant relationship. We concluded that the proportion of females in the board rooms doesnot necessarily affect firm performance as viewed with the use of ROE but other factorstogether with gender proportion exert a combined effect and these other factors are correlatedand therefore affect the performance of each other either positively or negatively. Thisscenario was therefore not realistic enough to establish a relationship between genderproportion and ROE.