Photodynamic therapy (PDT) of cancer typically involves systemic administration of tumor-localizing photosensitizers followed 48-72 h later by exposure to light of appropriate wavelengths. Knowledge about the distribution of photosensitizers in tissues is still fragmentary. In particular, little is known as to the detailed localization patterns of photosensitizers in neoplastic and normal tissues as well as the relationship between such patterns and the actual targets for the photosensitizing effect. This review focuses on ultrastructural features seen in treated cells and tumors. An attempt is made to correlate these findings with the subcellular/intratumoral localization pattern of the photosensitizers in tumor cell lines in vitro and in tumor models in vivo. Several subcellular sites are main targets of PDT with different sulfonated aluminum phthalocyanines (AIPcSn) in the human tumor cell line LOX. Nuclei are not among the primary targets. Overall, the ultrastructural changes correlate well with the data about the subcellular localization patterns for each analogue of AIPcSn in the same cell line. Similar findings are also obtained for the family of sulfonated mesotetraphenylporphines (TPPSn) in the NHIK 3025 cell line. The mechanisms involved in the killing of tumors by PDT seem to be a complex interplay between direct and indirect (via vascular damage) effects on neoplastic cells according to the intratumoral localization pattern of the applied dye. Several factors can affect the localization pattern of a drug, such as its chemical character, the mode of drug delivery, the time interval between drug administration and light exposure, and tumor type. Furthermore, whether local immune reactions (such as macrophages) and apoptosis (programmed cell death) are involved in the destruction of neoplastic cells by PDT in vivo is still an enigma. A general model for PDT-induced tumor destruction is suggested.