Tumour angiogenesis and the levels of plasminogen activator inhibitor type 1 (PAI-1) are both informative prognostic markers in breast cancer. In cell cultures and in animal model systems, PAI-1 has a proangiogenic effect. To evaluate the interrelationship of angiogenesis and the PAI-1 level in breast cancer, we have evaluated the prognostic value of those factors in a total of 228 patients with primary, unilateral, invasive breast cancer, evaluated at a median follow-up time of 12 years. Microvessels were immunohistochemically stained by antibodies against CD34 and quantitated by the Chalkley counting technique. The levels of PAI-1 and its target proteinase uPA in tumour extracts were analysed by ELISA. The Chalkley count was not correlated with the levels of uPA or PAI-1. High values of uPA, PAI-1, and Chalkley count were all significantly correlated with a shorter recurrence-free survival and overall survival. In the multivariate analysis, the uPA level did not show independent prognostic impact for any of the analysed end points. In contrast, the risk of recurrence was independently and significantly predicted by both the PAI-1 level and the Chalkley count, with a hazard ratio (95% CI) of 1.6 (1.01-2.69) and 1.4 (1.02-1.81), respectively. For overall survival, the Chalkley count, but not PAI-1, was of significant independent prognostic value. The risk of death was 1.7 (1.30-2.15) for Chalkley counts in the upper tertile compared to the lower one. We conclude that the PAI-1 level and the Chalkley count are independent prognostic markers for recurrence-free survival in patients with primary breast cancer, suggesting that the prognostic impact of PAI-1 is not only based on its involvement in angiogenesis.