Identification and treatment of involved axillary lymph nodes is important in the planning of strategies for adjuvant treatments of breast cancer. With the advent of the National Health Service Screening Programme, an increasing number of women with the disease are detected at an early stage, when the lymph nodes are not involved. In whom, therefore, is it necessary to carry out a formal axillary dissection? Are there accurate surrogates for lymph node involvement in the form of tumour markers or characteristics? This study, carried out on over 81 patients, examines the use of neural networks to predict the involvement of lymph nodes using readily available clinical and pathological data and also more specialized markers of possible prognostic significance. The study shows that neural networks are capable of providing strong indicators as to lymph node status using only basic measurements of the primary breast tumour. However, accuracy can be improved by the addition of less common markers.