Photodynamic therapy represents a new approach for the local control of cancers. It has recently been claimed that photodynamic therapy mediated by hematoporphyrin derivative (HPD) is selectively more efficient for killing leukemic cells than normal progenitors. To improve this effect, we studied the influence of hematoporphyrin dose, temperature during incubation and/or treatment, hematoporphyrin derivative incubation time, and fractionation of the argon laser light (488-514 nm) used for hematoporphyrin stimulation. Plating efficiency calculated after a 7-day period of growth on collagen gel medium showed a dose-dependent phototoxicity of HPD reaching 0.01% for normal hemopoietic progenitors and 0.001% for leukemic cells (dose = 12.5 micrograms/ml). The 10:1 ratio of normal hemopoietic progenitors to leukemic cells was also found to be the same or increased when temperature was 37 degrees C during incubation and 4 degrees C during laser irradiation. Similar results were also found when incubation time was varied from 75-120 min, or when laser irradiation dose was fractionated into 2 or 3 periods. The ratio of normal progenitors to leukemic cells reached 100:1 when 75 J/cm2 were fractionated into 3 periods after an incubation time of 120 min with 10 micrograms/ml HPD. Selectivity in photodynamic treatment seems to occur between normal hemopoietic progenitors and leukemic cells. The mechanism of this selectivity remains unclear, but experiments with the fractionated irradiation dose suggest that as in radiotherapy, better potentially lethal damage repair in normal cells could be a factor for selectivity in photodynamic therapy. Our results obtained with leukemic cells are fully in agreement with data in the literature concerning similar experimental models.