PurposeThe purpose of this study is to compare the use of diagnostic mammography, diagnostic mammography with ultrasound, and ultrasound alone in the evaluation of recalled non-calcified lesions from screening mammography with digital breast tomosynthesis (DBT). MethodsWe performed a retrospective review of recalled non-calcified lesions that included architectural distortion, asymmetry, focal asymmetry and mass from screening DBT from January 2014 to December 2016. Electronic health records were reviewed for imaging evaluations, findings and histopathology results. ResultsOf 266 non-calcified lesions in 247 women, masses were significantly more likely to be evaluated with ultrasound alone compared to diagnostic mammography with ultrasound than all other recalled finding types (OR = 7.63; 95 %CI [4.17–13.97]; p < 0.01). Architectural distortions were more likely to be assigned a BI-RADS 4 or 5 on the diagnostic evaluation than all other lesion types (OR = 7.71; 95 %CI [2.82–21.04]; p < 0.01). Masses were more likely to be true lesions (90 %; 111/124) compared to 40 % (23/57) of focal asymmetries (OR = 11.41; 95 %CI [4.05–32.14]; p < 0.01) and 24 % (16/67) of asymmetries (OR = 27.01; 95 %CI [9.41–77.48]; p < 0.01). No significant difference was noted in the malignant versus benign biopsy outcomes among the recalled DBT lesion types (p = 0.71). ConclusionRecalled masses from screening DBT were more likely to be worked up with ultrasound alone compared to all other non-calcified lesion types and more likely to represent true lesions on diagnostic evaluation. Recalled asymmetries and focal asymmetries were more likely to be worked up with either diagnostic mammography with ultrasound or diagnostic mammography alone.