Abstract Breast carcinoma is a heterogeneous disease with a prognosis that varies from excellent to very poor. Traditional tumour parameters and biological factors that are also predictive for treatment response are used in determining breast carcinoma prognosis and selecting appropriate treatment. Gelatinases MMP-2 and MMP-9 have been shown to associate with tumour progression. Their tissue inhibitors TIMP-1 and -2 are multifunctional molecules that have been suggested as prognostic markers in some previous reports. In the present work, the expression and prognostic value of gelatinases MMP-2 and MMP-9 and their tissue inhibitors TIMP-1 and -2 were assessed in primary breast carcinoma. The material consisted of a total of 416 patients. Tissue expression of TIMP-1 and -2 was analysed in a population of 203 patients using immunohistochemistry. Circulating gelatinases and their inhibitors were studied using ELISA in two different populations of 71 at preoperative state and 213 patients at pre- and postoperative state. High expression of TIMP-1 immunoreactive protein positively correlated with high histological grade of the tumour and associated with aggressive disease course in grade 2–3 subpopulation. High preoperative plasma TIMP-1 was prognostic for relapse in a modern patient series after a median follow-up time of 18 months. TIMP-1 as a continuous variable was prognostic in Cox regression univariate analysis, and was an independent prognostic variable superior to nodal status in multivariate analysis. High preoperative serum TIMP-1 was an independent prognostic variable for poor disease-specific survival, and TIMP-1 was found to maintain its prognostic value when assessed independently with different ELISA analyses, and was not very sensitive for preanalytical conditions. In addition, low circulating preoperative serum MMP-2 was observed to associate with high stage and positive nodal status in breast carcinoma. These results indicate that circulating TIMP-1 may be a potential new marker of worsened prognosis in breast carcinoma, although careful validation of assay platforms and identification of the sources of physiological variation are needed before it can be adopted into clinical decision-making.