The density profile of microvessels (MVD) has recently emerged as a prognostically independent morphological marker of various malignancies, including carcinoma of the breast. MVD-scoring may thus afford the pathologist the opportunity to identify subsets of early stage breast cancer patients, which may benefit from adjuvant therapy. Reproducibility studies are, however, mandatory. Fifty sections of invasive breast carcinoma, immunolabelled for Factor VIII-related antigen, were scored independently by two observers for the number of vessel-profiles (v-p) in the following manner: 1) 250 randomly selected high power fields (HPF) were scored by two analysts to evaluate the agreement of the counting per se; and 2) the hot spot(s), i.e. the zones considered the most vessel rich, in the 50 sections were identified and scored twice to evaluate the agreement on selecting hot spots. When the observers evaluated MVD in the very same 250 HPF a median deviation of 2 v-p (10%) was produced. The interobserver disparity was further accentuated when the participants independently had to identify the hot spots with a median deviation of 6 v-p (13%). Intraobserver variation was approximately equal to the interobserver discordance. Divergences were most conspicuous in fields with a complex vasculature. In conclusion the v-p scoring resulted in substantial inter- and intraobserver variation both in selecting hot spots and in performing the scoring per se. Alternative-scoring techniques, including strict stereological principles, should be considered in the morphological evaluation of angiogenesis.