BACKGROUND: Qunatitation of blood flow within the cardiac chambers has been tried using digitally recorded velocity information of color Doppler flow mapping (CDFM). However, the accuracy of CDFM has not been systematically assessed. The purpose of the study was to test the accuracy of digitally recorded velocity information obtained through CDFM. Method: Using the American Society of Echocardiography flow phantom, an in vitro steady flow experiment was performed. Eight different flow fields through the phantom (84, 139, 172, 192, 232, 237, 244, 436 mL/sec) were imaged by CDFM using Hewlett Packard Sonos 2500 and all data were stored in the magneto-optical disc. Multiple aliases were visible and 7 different Nyquist limits (17, 23, 27, 37, 42, 53, 58 cm/sec) were used for recording. A fully automatic computer algorithm was used to unwrap the aliases and to convert the CDFM to digital velocity. Among 9 consecutive values at each pixel along the centerline of the phantom, median value was chosen as a representative value, which was compared to the actual value. RESULTS: In the slow velocity region, CDFM tends to overestimate the true velocity, whereas in the fast velocity region, it underestimates. At the aliasing point of 2 times Nyquist limit, there is a marked turbulence and estimation of velocity is impossible;fter aliasing, velocity information of CDFM is totally unreliable. Velocity estimation of CDFM is the most accurate at the velocity region of the Nyquist limit and shows reasonable accuracy between 0.5 x Nyquist limit and 1.5 x Nyquist limit with the error of estimation less than 15%. Conclusion: Velocity information of CDFM is accurate in the limited region and is dependent on the Nyquist limit of each measurement;this can be explained by the effect of wall filter. Baseline-shift of CDFM should be limited to the region between 0.5 x Nyquist limit and 1.5 x Nyquist limit for accurate quantitation using velocity information of CDFM.