The aim of this study was to test the hypotheses that (a) microvascular density (MVD) measured in histological sections of resected non-small cell lung carcinomas is an index of angiogenesis and (b) the measurement of MVD in a single block is representative of the overall MVD of the tumour. MVD was quantitated in one block per specimen of 60 lung tumours and nine normal lung tissues, and in 47 blocks taken from different regions of four tumours. Blood vessels were stained with antibody to von Willebrand Factor and MVD was quantitated using two methods: average density throughout the section (a-MVD) and density in the most vascularized area or 'hot spot' (h-MVD). Similar h-MVD values were found in tumours and in normal bronchus, whereas a-MVD was greater in the latter (P < 0.01). When 47 blocks from four tumours were analysed, inter-tumour variation was significant (P < 0.001) in spite of significant intra-tumour variation. The highest MVD value was not necessarily found in the periphery of the tumour. The four tumours were ranked into either two or four tiers according to their overall MVD. In 50 random selections of one block per tumour, the correct ranking was achieved in 68-74% of cases with the two-tier ranking and in 6-16% of cases with the four-tier ranking (h-MVD and a-MVD values respectively). These results suggest that elevated MVD values do not necessarily represent angiogenesis in non-small cell lung carcinomas. When only one block per tumour is examined, the chance of obtaining an accurate estimate of the vascularity of that tumour may be lower than 68%.