Using a dynamic and symmetrical cardiac phantom different echocardiographic mathematical models (Simpson 7 slices, area-length method, Simpson 2 slices and method according to Teichholz) were compared. 9 different end-diastolic (EDV) and end-systolic (ESV) volumes, 9 different stroke volumes (SV) and ejection fractions (EF) were used. EDV and ESV varied between 39-298 ml; SV between 29-100 ml and EF between 14-46%. In addition 10 fixed volumes of the same shape were evaluated using the same echocardiographic mathematical models. While symmetrical fixed volumes can be assessed correctly (r = 0.97-0.98), apart from the formula according to Teichholz (r = 0.89, significant underestimation of volumes), the correlation coefficients decrease using a dynamic cardiac phantom. In the modification of Simpson with 7 slices the best correlation was found for all parameters (EDV: r = 0.93; ESV: r = 0.94; EF: r = 0.87; SV: r = 0.81). The biplane area-length method has no advantages over Simpson's rule with 2 slices in the short axis; for symmetrical models both methods are comparable, both having high correlation coefficients (for volumes r = 0.85 and r = 0.88; for EF 0.78 and 0.84). Using the method according to Teichholz symmetrical volumes can be well assessed (r = 0.90), for the determination of EF the correlation coefficient decreases to r = 0.65 and for stroke volume to 0.33, reflecting no significant correlation to the actual SV. Possible causes for a poorer correlation are discussed for moving objects as opposed to the fixed volumes.