AIM: To test whether immunoradiometric or immunohistochemical detection of lysosomal protease cathepsin D in breast cancer is more predictive of outcome. METHODS: Tumour tissues from 270 primary breast cancer patients were evaluated for the expression of cathepsin D using immunohistochemistry (IH; paraffin embedded tissues) and an immunoradiometric assay (IRMA; cytosol from frozen tissues). Immunohistochemical scores were based on immunoreaction in tumour cells and tumour associated macrophages. RESULTS: IRMA values (cut off 40 fmol/mg cell protein) correlated significantly with IH values. Recorded incidences of positive immunoreaction in tumour cells using two different cut off values were 52% and 35%, respectively. Macrophages stained positive in 31% of tissues. Combined evaluation of tumour cells and macrophages resulted in positivity rates of 59% and 48%, respectively. Node status was the only variable found to correlate with cathepsin D expression. IH results correlated significantly with clinical outcome (median observation time 68 months) in node negative patients (n = 120) but not in node positive patients (n = 145). Cathepsin D positivity as measured by IRMA was not related to clinical outcome in either group. On multivariate analysis in the node negative group, IH detection of cathepsin D appeared to be the only independent factor indicating prognosis. For node positive patients, tumour grade, size, and receptor status were of prognostic relevance. CONCLUSIONS: Because of the simple methodology and the minimal amount of tissue used for analysis, immunohistochemistry was preferred to immunoradiometry for cathepsin D measurement; it also provided more predictive data with respect to prognosis.