The most commonly used system in Europe for breast carcinoma was developed by Scarff, Bloom and Richardson (SBR). It was recently modified by Elston and Ellis and significant improvement in reproducibility has been shown by using precise grading guidelines. This study investigated whether the use of this new grade (defined as the Nottingham grade, NG) would improve the prognostic stratification of patients. The respective prognostic value of the two grading schemes was compared in a retrospective series of 825 patients uniformly treated for a small invasive breast carcinoma and followed for a median of 6 years. Univariate and multivariate analysis showed that both histological grades were strongly correlated to overall and metastasis free survival. We have separately analysed the prognostic value of each of the three components used to assess the two grading systems and found that the mitotic index was the only significant prognostic factor for 5 year survival. Univariate analysis showed the count to be more discriminant in the NG scheme (p = 0.0006) than in the SBR scheme (p = 0.04). However, in univariate and multivariate analysis, the prognostic value of the global NG was not significantly better than SBR grade. This may be related, in part, to an uneven distribution of cases reflected by a much lower number of cases with a high mitotic index in the NG system (2%) than in the SBR system (10%). Our study emphasizes the importance of the mitotic count in assessing the prognosis of breast cancers and indicates that the factors which condition this count (tissue processing, microscopic observation, threshold) must be well standardized and controlled.