Anomalies of the coronary ostia can have severe consequences. To provide a screening solution, automated ostia detection based on single breath-hold coronary MRA scans is presented. The aorta is segmented in the data sets to serve as an orientation point and vesselness enhancing filters are applied. Searching the aorta surface for high vessel responses by a ray-tracing procedure yields information about the position of the coronary ostia. The proposed approach was successfully validated in 10 volunteers with an average deviation of $$$7.6 \pm 1.0$$$° in angular and $$$1.2 \pm 0.58$$$ mm in superior-inferior direction.