The aim of the present study was to obtain the precision of flow measurement in breath-hold segmented k-space flow sequences. The results are based on studies of pulsatile flow in a phantom tube. The ultimate purpose is to use these sequences to measure coronary flow. In abdominal and cardiothoracic magnetic resonance imaging the image quality is degraded due to respiratory motion. In the segmented k-space acquisition method, one obtains many phase-encoding steps or views per cardiac phase. This shortens imaging time in the order of phase-encoding lines and makes it possible to image in a single breath-hold, thereby eliminating respiratory artefacts and improving edge detection. With breath-hold multiframe cine flow images it is possible to evaluate flow in all abdominal and cardiothoracic areas, including the coronary arteries. Our study shows that velocity curves shift in time when the number of k-space ky-lines per segment (LPS) are varied; this shift is linear as a function of LPS. The mean velocity Vmean in the center of mass of the pulsatile peak is constant (Vmean = 40.1 +/- 2.9 cm/s) and time t = -10.1 x LPS + 268 (r = 0.993, p < 0.0001). Correlation between theoretical and experimental flow curves is also linear as a function of LPS: C = -0.977 * LPS (r = 0.987, p < 0.0001). It is concluded that velocity curves move with LPS and are smoothed when the breath-hold velocity mapping is used. The more LPS is gathered the more inaccurate results are. LPS 7 or more cannot be considered clinically relevant.