Abstract: It is impossible to predict the occurrence of large earthquakes by using experimental results of rock meta-instability stage before the rock instability and rupture. This judgment is supported by the following arguments: ① The significant difference between the large and small earthquakes is that, for the large ones, there is a long period （tens or hundreds of seconds） fault dynamic fracture after the initial fracture, while for small ones there isn’t. At present we can’t predict when the dynamic fracture will stop, and how large the final earthquake fault will be. ② If we prefer to predict tectonic earthquakes based on the rock meta-instability observation, then we can only predict the occurrence of initial fracture of an earthquake, rather than how large it will be. ③ By inverting the seismic wave observations we can understand the process of fault dynamic fracture. At present, the fault dynamics can’t give a practical theoretical model of fault dynamic fracture, therefore prediction cannot be made.