Color velocity imaging (CVI) is a new non-Doppler ultrasound technique for vascular color flow imaging. Using information contained in the two-dimensional B-mode, gray-scale image to determine velocity, CVI offers potential advantages over Doppler color flow imaging methods. In order to be used clinically, velocity determination with CVI must be validated by other current methods. A Doppler string phantom was studied with a Philips CVI ultrasound system. Velocity measurements were obtained by both CVI and duplex Doppler spectral analysis for constant string speeds from 10 to 200 cm/sec, at intervals of 10 cm/sec. Twenty separate estimates were obtained with each method, at each string speed. Linear regression assessed the relationship between estimated and actual string velocities, with CVI and spectral Doppler analysis yielding highly valid results (CVI = -0.713 + 1.000997 x phantom; r 2 = 0.9979). At all string speeds tested, the averaged estimated and the actual velocities for both methods were within the 95% confidence estimates. The range for the CVI 95% confidence limits from the regression line varied from +/-1.07 cm/sec at the lowest speed of 10 cm/sec (11.6%) to +/-7.72 cm/sec at 200 cm/sec (3.87%). Based on in vitro testing, CVI is as accurate as Doppler spectral analysis for the estimation of flow velocity.