The objective of this study was to compare the accuracy of strain elastography (SE), 3-D ultrasound (US), B-mode US with compound imaging (CI) and B-mode US without compound imaging for lesion sizing in breast cancer. The prospective study included 93 patients with invasive breast cancer. The largest tumor diameters measured by B-mode US, B-mode US with CI, SE and 3-D US were compared in Bland–Altman plots versus pathology as reference. A general linear model repeated measures (GLM Rep) was applied to investigate factors influencing tumor sizing. All methods underestimated pathologic size, with SE (−0.08 ± 7.7 mm) and 3-D US (−1.4 ± 6.5 mm) having the smallest mean differences from pathology. Bland–Altman plots revealed that B-mode US, B-mode US with CI and 3-D US systematically underestimated large tumor sizes, and only SE was technically comparable to pathology. The study indicates that sonographic underestimation of tumor size occurs mainly in tumors >20 mm; in this subgroup, SE is superior to other ultrasound methods.