Seismic tomography is extensively used as a means of understanding structures, but how far can the resulting images be trusted? What influence comes form the inversion technique? Does loss of amplitude of a wavespeed anomaly come from a real variation or merely lower attainable resolution? What reliable information can be extracted even where resolution is variable? These issues are particularly important when considering features such as possible "tears" in the subducting slab that can influence the propagation of high frequency waves from deep earthquakes.