A p-type semiconductor nanowire transistor is formed on the first semiconductor nanowire and an n-type semiconductor nanowire transistor is formed on the second semiconductor nanowire. The first and second semiconductor nanowires have a rectangular cross-sectional area with different width-to-height ratios. The type of semiconductor nanowires for each semiconductor nanowire transistor is selected such that top and bottom surfaces provide a greater on-current per unit width than sidewall surfaces in a semiconductor nanowire having a greater width-to-height ratio, while sidewall surfaces provide a greater on-current per unit width than top and bottom surfaces in the other semiconductor nanowire having a lesser width-to-height ratio. Different types of stress-generating material layers may be formed on the first and second semiconductor nanowire transistors to provide opposite types of stress, which may be employed to enhance the on-current of the first and second semiconductor nanowire transistors.