OBJECTIVE To investigate the general applicability and interobserver variability of ultrasonographic (US) features in differentiating benign from malignant solid breast masses. METHODS One hundred sixty-two consecutive solid masses with a tissue diagnosis were reviewed. Three radiologists reviewed the masses without knowledge of clinical history or histologic examination results. RESULTS US features that most reliably characterize masses as benign were a round or oval shape (67 of 71 [94%] were benign), circumscribed margins (95 of 104 [91%] were benign), and a width-to-anteroposterior (AP) dimension ratio greater than 1.4 (82 of 92 [89%] were benign). Features that characterize masses as malignant included irregular shape (19 of 31 [61%] were malignant), microlobulated (four of six [67%] were malignant) or spiculated (two of three [67%] were malignant) margins, and width-to-AP dimension ratio of 1.4 or less (28 of 70 [40%] were malignant). If the three most reliable criteria had been strictly applied by each radiologist, the overall cancer biopsy yield would have increased (from 23% to 39%) by 16%. When US images and mammograms were available, the increase in biopsy yield contributed by US was not statistically significant (2%, P = .73). However, in independent reviews, one to three reviewers interpreted four carcinomas as benign at US. CONCLUSIONS The data confirm that certain US features can help differentiate benign from malignant masses. However, practice and interpreter variability should be further explored before these criteria are generally applied to defer biopsy of solid masses.