Seventy-one cases that had resulted borderline for HER-2 protein expression at conventional immunohistochemical assay (2+) were assessed for HER-2 gene amplification by real-time PCR and by FISH in accordance with the manufacturer's recommendations (gene amplification with ratio >or=2 in both methods). Thirty-three out of 71 cases (47%) resulted amplified at real-time PCR analysis, whereas 15 cases resulted positive at FISH (21%). Apparently, PCR was more sensitive than FISH in HER-2 determination, only 10 cases resulting amplified in both tests. When the mean ratio value obtained in all PCR experiments was adopted as threshold in determining HER-2 gene amplification, the apparent sensitivity of PCR was reduced but correlation between PCR and FISH results was dramatically increased. Furthermore, when the mean PCR ratio value observed in the FISH-positive group was chosen as threshold, the best agreement between PCR and FISH results was achieved. Therefore, we found that the proposed threshold ratio value of >or=2 is not accurate in separating HER-2 amplified and non-amplified cases. We suggest that the threshold ratio value in PCR tests should be determined in each laboratory using FISH controlled cases. Finally, above certain in-lab generated threshold values, PCR might be proposed as a highly predictive positive test in HER-2 assessment.