Echocardiography is an established method to estimate left-ventricular mass (LVM) in mice. Accuracy is determined by cardiac size and morphology and influenced by mathematical models. We investigated accuracy of three common algorithms in three early developmental stages. High-resolution echocardiography was performed in 35 C57/BL6-mice. Therefore, two-dimensional-guided M-mode echocardiography and parasternal short- and long-axis views in B-mode were obtained. LVM was assessed in vivo applying Penn (P), Area Length (AL), and Truncated Ellipsoid (TE) algorithms and validated with histomorphometry. Regression analysis of all mice showed fair estimation of LVM assessed with M-mode-based Penn algorithm (y = 0.6*x - 0.12, r: 0.71). In contrast two-dimensional assessment of LVM revealed close linear relationship with histomorphometry (y(AL)= 1.21*x - 12.1, r: 0.88, y(TE)= 1.38*x - 2.88, r: 0.86). Bias was lowest for LVM-AL at diastole underestimating 3.2%. In concordance with the summarized data, LVM-P revealed lower regression coefficients and significant underestimation in all three subgroups. Small hearts (<50 mg, n = 12) correlated best with LVM-AL at systole. Hearts of adolescent (50-75 mg, n = 13) and adult (75-100 mg, n = 10) mice revealed close linear relationship with LVM-AL and LVM-TE at diastole. Echocardiographic assessment of LVM is feasible in hearts weighting less than 50 mg and can be estimated best in systole. Hearts weighting more than 50 mg are estimated most accurately by means of LVM-AL at diastole.