In cine cardiac magnetic resonance imaging (MRI) studies, for any preset imaging parameters the number of phases per cardiac cycle for a single slice is proportional to breath-hold duration. We investigated the relationship between the accuracy of measurement of left ventricular (LV) end-diastolic and end-systolic volumes (EDV and ESV, respectively), mass and ejection fraction (EF), and the number of phases acquired per cardiac cycle. Twelve adult volunteers underwent cardiac MRI and five complete LV functional studies were obtained with 8, 11, 14, 17, and 20 phases per cardiac cycle. We calculated LV volumes, EF, and mass for each acquisition, and compared them using the 20-phase acquisition as the reference standard. The scan duration was proportional to the number of phases acquired. There was a systematic underestimation of LV, EDV, and EF, with decreasing number of phases. Differences from the reference standard became significant for the 8-phase acquisition (p<0.05). Subgroup analysis showed that only those with slower heart rates (<65/min) had significant differences in EDV, but not in EF, for the 8-phase acquisition. For those with faster heart rates, no differences were detected between the different acquisitions. There were no significant differences between all acquisitions for the LV ESV and mass. We conclude that at least 11 phases per cardiac cycle are needed to maintain accuracy for cine cardiac MRI studies. Decreasing the number of phases per cardiac cycle beyond this cutoff may introduce significant error of measurement, particularly for the left ventricular EDV and EF and especially for those with bradycardia, and should be avoided.