Recent investigations demonstrate a strong potential for Cobalt-60 (Co-60)-based tomotherapy. Reported work suggests that Co-60-based tomotherapy offers a clinically and commercially viable alternative to megavoltage x-ray-based tomotherapy. Tomotherapy applications use a combination of intensity-modulated fan beams to deliver highly conformal radiotherapy. However, conventional Co-60 units are designed to give large uniform rectangular fields using an isotropic radioactive source in a cylindrical geometry. Such cylindrical source geometry likely provides a sub-optimal use of the radioactivity within the source volume for tomotherapy applications due to a significant loss of radiated energy outside the fan beam collimation system. To investigate a more efficient source geometry suitable for Co-60 tomotherapy applications, a computer code was written to model an isotropic source in a 6-faced polyhedron geometry such as cube, parallelepiped, prism and truncated pyramid. This code was integrated with the existing EGSnrc/BEAMnrc Monte Carlo (MC) code. The integrated source code was thoroughly tested, validated and used to investigate the energy spectra, radiation output and self-shielding properties of various rectangular-shaped (RS) Co-60 sources. Fan beam dose profiles were calculated for various cylindrical and RS Co-60 sources for a range of source-to-axis distances (SAD), multi-leaf collimator-to-isocentre distances (CID) and modified collimator systems. Fringe and penumbra distances were analysed for the simulated dose profiles. Our results demonstrate that clinically acceptable fringe and penumbra distances can be achieved by a careful selection of SAD, CID, source shape and dimensions and modified collimator system. Significant overall gain in radiation output of the 20 x 1 cm(2) fan beams can be achieved by an optimal selection of the source geometry for a given active volume of Co-60. The overall gain includes the effects of change in packing density (accounting for self-absorption) and change in source shape.