For deep events and epicentral distances around 75° the Bulletins of the International Seismological Centre (BISC) contain many secondary arrivals between P and PcP, the reflection from the core-mantle boundary (CMB), which cannot be explained by current, spherical symmetric Earth models. Assuming that these arrivals in the P-coda are reflections from P-velocity anomalies in the lower mantle, we calculate the associated reflection points (geographical coordinates and distance from the CMB) and map both strength (relative density of reflection points in bins in the lower mantle) and distance of the reflectors from the CMB on a global scale. The regions with these anomalous bins cover 3% of the 50% of the lowermost mantle that can be analysed with this data set. The most significant regions with anomalous bins (values in parentheses give a preliminary estimate of the distance of the reflector from the CMB) are under the North Atlantic Ridge (250 km), northern Central Siberia (230 km), Japan (180 km), the central Mid-Atlantic Ridge (310 and 260 km) and off the coast of Ecuador (340 km). The presence of reflectors in one of these regions has already been confirmed by detailed seismic studies using waveform data from arrays and networks. A slight (2–3%) P-velocity contrast is sufficient to produce measurable reflection amplitudes in the distance range of 70–80°. Ten percent of the lower mantle sampled by this study is covered by bins with no or very few secondary arrivals in the P-coda. The most significant regions with such a ‘normal’ lower mantle are under Central America, the Caroline and Marshall Islands in the eastern Pacific, the area of Hawaii and the Easter Islands and the northern Atlantic. Identification of the regions with anomalous bins serves to locate the areas of the lowermost mantle which should be analysed further by using the full information content of seismograms. The study of the travel time, the amplitude and the form of the wavefield, recorded preferably at seismic arrays/dense networks, will allow the determination of the anomalous velocity structure of the core-mantle transition zone in more detail. The most promising areas for such a study are given here, but further anomalies are likely to exist.