There is no information available on the relation between response to chemotherapy and the high-risk phenotype assessed by uPA and/or PAI-1. The clinical situation of neoadjuvant chemotherapy provides a means of rapidly assessing the sensitivity of the primary tumour to cytotoxic drug regimens. The goal of the study was to assess prospectively the predictive value of PAI-1 for response to first-line chemotherapy. PAI-1 concentration was measured on hypertonic cytosolic extracts (0.4 M potassium chloride) by ELISA before chemotherapy on a drill biopsy sample of the tumour in 69 T2 and T3 breast cancer patients (median age 46 years). Oestrogen receptor (ER) (51% ER+), progesterone receptor (PR) (58% PR+), S-phase (median 4.0%) and ploidy were also assessed in the majority of cases. The clinical response to treatment was evaluated after four cycles of FAC or FEC regimen (5-fluorouracil, epidoxorubicin or doxorubicin and cyclophosphamide) (one cycle every 4th week). PAI-1 could be assayed in 29 post-chemotherapy surgical samples. The objective response rate (complete response plus partial response) was 59% (41 out of 69). PAI-1 expressed as gram of tissue (range 19-2370 ng g(-1) tissue) was highly correlated (r = 0.98) to PAI-1 expressed as mg protein (range 0.5-68 ng mg(-1) protein). No correlation between PAI-1 level and response could be observed, with any cut-off. The post- and pre-chemotherapy PAI-1 levels were correlated (r = 0.66). Of all biological parameters, only high S-phase (cut-off 5%) was slightly correlated (chi2 = 3.91, P = 0.05) to response. These data suggest that PAI-1 is not a predictive marker of response to chemotherapy in breast cancer and that its level is not altered by neoadjuvant chemotherapy.