Core biopsy is currently used as the initial diagnostic procedure in breast cancer. Prognostic information is required before decisions on subsequent therapy are possible. The aim of this study was to compare prognostic data observed in ultrasound-guided core biopsies and in surgical specimens. We conducted a retrospective study of a mean of 1.2 biopsies from a total of 110 consecutive malignant breast masses. Histological tumour type and grade, presence of a ductal carcinoma in situ (DCIS) component and hormonal status were evaluated. Our results showed excellent sensitivity (100%) of core biopsy for the diagnosis of breast cancer and for the assessment of tumour type (concordance of 73.6%). We also reached a reasonable level of agreement between core biopsy and surgical excision specimens for histological grading (73.1%). Underestimation of grading on core biopsy was due predominantly to underscoring of mitotic counts (32.3%). The presence of DCIS associated with infiltrative carcinomas was widely underestimated in biopsy specimens. There were good levels of agreement for oestrogen receptors and for progesterone receptors: 90.3% and 89.3%, respectively. Diagnoses were made on the basis of a small number of ultrasound-guided core biopsy specimens taken from breast masses. Grade and hormonal status were reliably evaluated.