Two human pancreatic cell lines, MIA PaCa 2 and Capan 2, were treated by photodynamic therapy in vitro with Photophrin (0.01-25 micrograms/ml; 24 h) and then light (1-50 J/cm2; lambda = 630 nm). The following model was fit to 6 datasets with weighted nonlinear regression: [sequence: see text] The symbols are: E, cell growth; Econ, control growth in the absence of the combination; B, background signal; m, slope parameter; gamma, interaction parameter; D, concentration of Photofrin; L, light dose; F, fraction of Photofrin not photobleached by the light dose; k, k1, k2, bleaching parameters; A, distribution parameter for biexponential bleaching equation. Simple reciprocity of photosensitizer concentration and light dose was not found; compensation for photobleaching was critical. MIA PaCa2 required the monoexponential bleaching factor, whereas Capan 2 required the biexponential bleaching factor. The greater photosensitivity of MIA PaCa2 over Capan 2 can be best explained not by differences in the interaction parameter but rather by differences in the photobleaching pattern and rate. It may be possible to further enhance the selectivity of photodynamic therapy if differences in photobleaching between different cell types can be exploited by adequate dosimetry.