Dosimetry in photodynamic therapy as currently practiced is empirical in that it does not account for optical properties of the target lesion. However, since light attenuation in tissue is unpredictable, measurements of optical properties are needed to ensure optimal light dose delivery. Further improvements in the uniformity of light dose distribution in tumors can be afforded by implanting multiple light sources. A technique is described in which the use of multiple cylindrical sources was combined with measurements of light energy fluence rate in the tumor. Six sources were placed within translucent plastic needles, which were inserted into tumors in a parallel array. Tumor attenuation characteristics were measured by placing a miniature light detector in one needle, while illuminating a cylindrical source in another, nearby, needle. This process was repeated for different needle pairs. In one postmortem and two in vivo tumors the absorption coefficient, transport scattering coefficient and penetration depth ranged from 0.56-0.81 cm-1, 9.4-15.2 cm-1 and 1.7-2.3 mm, respectively. Apparent penetration depths for in vivo tumors changed with time, during experiments. Predictions of dosimetry were generally consistent with direct measurements of light in tumors. Somewhat better agreement was observed in an optical phantom.