Breast cancer continues to be the most common cause of cancer deaths among women. Early detection of breast cancer is vital to improve its prognosis. Digital Mammography currently offers the best control strategy for early detection of breast cancer. The research work in this paper investigates the significance of neural association of microcalcification patterns for their classification in digital mammogram. The proposed technique explores the auto-associative abilities of a neural network approach to regenerate the composite of learned patterns most consistent with new information, which uniquely signifies each class of input patterns, and improves the overall classification. It uses two types of features: computer extracted (grey level based statistical) features from mammogram; and human extracted (radiologists’ interpretation) features to classify different types of breast abnormalities. On testing dataset it attained 90.5% classification rate for calcification cases and 89.7% classification rate for mass cases.