A proportion of women thought to have ductal carcinoma in situ (DCIS) on mammography and a core biopsy showing DCIS only, in fact have an invasive focus on surgical excision. This study aims to identify the percentage of such patients who harbour an invasive focus and to ascertain features which can predict the presence of invasion. 140 patients had a core biopsy diagnosis of DCIS without invasion. All patients had their core biopsy graded and mammography was performed on 128 patients. Mammographic findings were classified by a radiologist blinded to the surgical findings into normal, mass/distortion or microcalcification. The extent of the microcalcifications was measured. The core biopsies were graded into high, intermediate or low grade DCIS groups. The core biopsy and radiological findings were compared to see if they could predict the presence of invasive disease at surgical excision. Of the 140 patients, 61 (44%) had an invasive focus. 8 (47%) of 17 patients with normal mammography had an invasive focus. 4 (36%) of 11 patients with a mammographic mass had evidence of invasion. Of the 100 patients with mammographic microcalcifications 48 (48%) had an invasive focus. In the 10 patients with low grade DCIS on core biopsy, 3 (30%) had an invasive focus. Comparative studies in patients with intermediate and high grade DCIS, were 7 of 18 (39%) and 51 of 112 (46%), respectively. Thus, 44% of women thought to have DCIS only on preoperative investigation had an invasive focus. In contrast to previous expressed opinions, neither mammography or grade were predictive. We have not identified any factor capable of predicting a higher likelihood of an invasive focus.