Increased efficacy of light and laser applications in medicine is achieved by accurate light dosimetry. A minimally invasive technique for the determination of the optical coefficients of tissue involves interstitial measurements of the local fluence rate at two or more points in the tissue using isotropic, fibre optic detectors and application of a diffusion model of light propagation. The diffusion models assume simple, homogeneous tissue geometries, possibly oversimplifying the effect of tissue heterogeneities and boundaries. The primary goals of this study were to investigate the influence of realistic finite geometries on the fluence rate distribution and to quantify the systematic errors in the derived optical properties. A Monte Carlo model was developed to predict the fluence rate distribution in any plane of interest in a medium and was verified by comparison with diffusion theory solutions for simple geometries. Fluence rate measurements were made in optically infinite and semi-infinite phantoms for a wide range of optical properties and it was determined that the optical coefficients were derived accurately for phantoms with ueff> 0.2 mm-1 and 2 < ut'<10 mm-1. Measurements were also made in finite spherical volumes with absorbing (Rd = 0.35) and diffuse reflecting (Rd =0.85) boundaries for three optical phantoms and comparisons of the experimental fluence rates with the predictions of the finite volume Monte Carlo model are presented. Boundary effects were observed to be significant within 4 transport mean free paths (mfp') of the boundary. The optical coefficients were derived by applying a diffusion solution for an infinite medium and it was determined that within 2 mfp' of the boundary, the derived ua was overestimated by 40% and underestimated by 20% for the absorbing and reflecting boundaries, respectively. / Thesis / Master of Science (MS)