Summary A method is presented in this report for concurrent analysis of vascular architecture, blood perfusion and proliferation characteristics in whole-tumour cross-sections of human larynx carcinoma and glioblastoma xenografts. Tumours were implanted subcutaneously in nude mice. After i.v. injection with Hoechst 33342 and bromodeoxyuridine (BrdUrd) as perfusion and proliferation markers, animals were killed. An antiendothelial antibody (9F1) was used to delineate vascular structures. Cross-sections were analysed by a multistep immune staining and a computer-controlled microscope scanning method. Each tumour section was stained and scanned four times (Hoechst, 9F1, BrdUrd and Fast Blue for all nuclei). When these images were combined, vasculature, perfusion and proliferation parameters were analysed. The labelling index (LI) was defined as the ratio of the BrdUrd-labelled area to the total nuclear area. The LI based on manual counting and the LI calculated by flow cytometry (FCM) were in good agreement with the LI based on surface analysis. LI decreased at increasing distance from its nearest vessel. In the vicinity of perfused vessels, the LI was 30-70% higher than near non-perfused vessels. This method shows that both vasculature/perfusion and proliferation characteristics can be measured in the same whole-tumour section in a semiautomatic way. This could be applied in clinical practice to identify combined human tumour characteristics that predict for a favourable response to treatment modifications.Keywords: squamous cell carcinoma; glioblastoma; image analysis; proliferation; vasculature; perfusion The radiation response of tumours is determined by several wellrecognized factors, including intrinsic radio sensitivity, cell kinetics and the degree of tumour oxygenation and perfusion. There is increasing recognition that a combination of these mechanisms may be responsible for treatment failure in certain tumour types.The prognostic relevance of intrinsic radio sensitivity, measured by clonogenic assays, has been demonstrated for cancer of the uterine cervix (Levine et al, 1995) and for head and neck cancer (Girinsky et al, 1993).Another cause for radiation treatment failure is tumour cell repopulation, which compensates for radiation-induced cell kill. The longer the overall treatment time of fractionated radiation treatments, the greater the opportunity for tumour cell repopulation. Withers et al (1988) have reviewed 59 clinical studies on head and neck cancer and have demonstrated that the outcome was worse with longer treatment schedules. Preliminary results from four randomized studies in head and neck cancer and one in bronchus carcinoma demonstrate that tumour control rates can be improved with shortened radiation treatment schedules delivering two or more radiation fractions per day relative to once-a-day treatments with conventional radiotherapy (Ang et al, 1996;Horiot et al, 1996;Overgaard et al, 1996;Saunders, 1996).Begg et al (1990) measured the potential doubling time (T 0t) in biopsies of hea...