In primary, operable breast cancer, the Nottingham Prognostic Index (NPI) based on tumour size, lymph node stage and histological grade can identify three prognostic groups (PGs) with 10-year survival rates of 83%, 52%, and 13%. With the aim of defining a subset of patients having so good prognosis that adjuvant therapy can be withhold, the NPI was applied to a Danish population-based study group comprising 9,149 patients. As opposed to the British study, we used conventional axillary lymph-node staging. Histological grading was in both studies done by means of a similar slight modification of the Bloom and Richardson procedure, but in the Danish study only ductal carcinomas were graded. The 10-year crude survival was 68.1% for 4,791 patients with tumour size < or = 2 cm and 70.0% for 2,900 patients with grade I tumours. For 4,761 node-negative patients, the 10-year survival was also 70.0%, the expected survival being 89.3%. The relative mortality (observed:expected) was even at 10 years 2.1 demonstrating that more than 10 years observation time is necessary to estimate cumulated mortality. By application of the NPI, the Danish good PG comprising 27.3% of the patients had a 10-year survival of 79.0%. Thus, the index defined a subset with better survival than could be defined individually by each of its three components, but it did not succeed in defining a subset with survival similar to the expected; additional prognostic factors are therefore needed. The somewhat poorer survival of the Danish good PG may be ascribed to the British inclusion of non-ductal carcinomas, to interobserver variation present only in the Danish study, and to poorer expected survival of the Danish patients. The 10-year survival of the Danish moderate PG and poor PG was 56% and 25%, respectively. These improved survival rates are attributed to the administration of adjuvant therapies. There were virtually no node-positive patients in the good PG and no node-negative patients in the poor PG. Patients should therefore still be stratified initially by lymph-node status, but tumour size and histological grade are significant prognostic factors primarily within the node-negative group, and they should be included in future prognostication procedures.