In cases of wire-guided excision of non-palpable breast cancer (WGE), data concerning the determinants and correlations between radiologic and histologic margins and residual cancer in re-excisions are sparse. A total of 21 variables in 66 WGE followed by 49 re-excisions were prospectively analyzed. In multivariate analysis, only large mammographic lesions were clearly related to positive margins in specimen radiography (P<0.05). Multifocality (P<0.001), large pathologic size (P<0.05) and superficial excision (P<0.05) were related to positive histologic margins and multifocality (P=0.001) to residual disease in re-excisions. The sensitivity, specificity and positive predictive values of specimen radiography for predicting histologic margins were 33%, 79% and 53%, and those for predicting residual disease 30%, 80% and 38%, respectively. The ability of histologic margins to predict residual disease was 91%, 58% and 38%, respectively. In WGE, large mammographic lesions carry a significant risk for radiologically incomplete excision, while pathologically large and multifocal tumors may be histologically incompletely excised, especially if the excision does not extend down to the pectoral fascia. The excision sites of multifocal tumors should be re-excised because of the considerable risk of residual disease. The radiologic and histologic margins of the specimen may be misleading.