PURPOSE To test the sub-second imaging achieved with ViewRay for accuracy and ability to track real-time target motion during treatment delivery. METHODS A motion phantom incorporating MR compatibility was designed using the QUASAR Respiratory Motion Phantom. A silicon di-electric gel poured into a cylindrical shape was used as the target, connected to the motor via a rod. The target was driven with a sinusoidal trajectory of 20 mm amplitude. The cycle length was varied from 3 to 9 seconds per cycle. Images were acquired in the sagittal plane every 0.25 seconds and processed in MATLAB. Morphological-based automatic contouring identified the target through intensity thresholds and smoothing via erosion and dilation on each frame. The trajectory, y(n), was plotted using the centroid of each contour as an estimate of position and was fitted to a sinusoidal curve with four fitting parameters. The resultant amplitude, angular frequency, and cross-sectional area were compared with the expected quantity. Motion tracking was also tested in canine patients. Both automatic and manual contouring on each frame was used to obtain a trajectory. RESULTS The average measured amplitude over each trajectory was 20.3±0.1mm, within 2% of the known amplitude. The expected angular frequency, calculated from 2 π /T, agreed with the measured value within 2% for each trajectory. The average recorded cross-sectional area was 3.9±0.2 cm². which matches with the expected 3.9 cm². Finally, the cine images from the canine study had sufficient image quality to quantify motion and track target trajectories. CONCLUSION Over the range of cycle lengths representing a wide range of breathing rates in patients, very good agreement was observed between the expected and measured target trajectory. Clinically relevant quantitative motion data can be extracted from the cine images.