Photodynamic therapy (PDT) is often thought to be able to effect selective tumour necrosis. This therapeutic selectivity, based on transient differences in tumour: normal tissue photosensitizer concentration ratios, is rarely useful clinically in extracranial tumours, although PDT itself may be of value by virtue of the nature of the damage produced and healing of normal tissue by regeneration. This report describes the effects of PDT on normal pancreas and chemically induced pancreatic cancers in the hamster, where a different mechanism of selective necrosis may be seen. Photosensitizer distribution in normal and neoplastic pancreas was studied by chemical extraction and fluorescence microscopy. Correlation of distribution studies with necrosis produced by PDT shows that the photodynamic dose (product of tissue concentration of sensitizer and light dose) threshold for damage is seven times as high for normal pancreas as for pancreatic cancer. Tumour necrosis extended to the point where tumour was invading normal areas without damaging the normal tissue. In rat colonic cancer, photodynamic dose thresholds in tumour and normal tissue are similar and so such marked selectivity of necrosis is not possible. The reason for this selectivity in the pancreas is not clear, but recent evidence has suggested a difference in response to PDT between normal and neoplastic pancreatic cell lines and the presence of a singlet oxygen scavenger in normal pancreas is postulated. Furthermore, the present fluorescence microscopy studies suggest that tumour stroma contains the highest level of photosensitizer and thus receives the highest photodynamic dose during PDT. These results suggest a possible role for PDT in treating small pancreatic tumours or as an adjuvant to other techniques, such as surgery, that reduce the main bulk of tumours localized to the pancreas.