For patients with nipple discharge (ND), surgical duct excision is often required to exclude underlying malignancy. Our objective was to define clinical predictors of malignancy and examine the utility of common preoperative studies.We retrospectively identified 475 patients presenting with a chief complaint of ND from 1995 to 2005; 416 (88%) were eligible for review.Following standard evaluation (clinical breast examiation/mammogram/ultrasound), 129 of 416 (31%) were considered to have physiological ND and were managed expectantly, whereas 287 of 416 (69%) underwent further evaluation (cytology/ductography/magnetic resonance imaging) followed by biopsy ± surgery. Clinical features associated with pathological ND included bloody ND (adjusted odds ratio 3.7) and spontaneous ND (adjusted OR 3.2). Biopsy/surgery identified a causative lesion in 259 of 287 (90%), of which 37% were either malignant (n = 65) or high-risk (n = 30) lesions. The sole clinical predictor of malignant/high-risk lesion was a palpable mass (adjusted odds ratio 4.3). Preoperative evaluation identified 76 of 95 (80%) malignant/high-risk lesions, whereas 19 of 95 (20%) were identified by duct excision alone.Although clinical stratification alone reliably identified patients with pathological ND, neither the clinical characteristics nor preoperative studies can reliably distinguish between benign and malignant pathology. Surgical duct excision remains the gold standard to exclude underlying malignancy.