An analysis of specific experimental problems encountered in the testing of concrete and concrete-like materials under impact loading, using the split Hopkinson pressure bar test (SHPB) or the so-called direct impact Hopkinson bar test (DIHB), is presented in this paper. On the one hand, in data processing, it is shown that recent improvements such as wave dispersion correction and exact time shifting are indispensable to obtain accurate measurements in those tests. On the other hand, the conventional analyses of those two kinds of tests are investigated with the aid of transient numerical simulations. It is concluded that only one of the conventionally used formula offers a good estimate of stress-strain curve for concrete-like materials. It is also illustrated that the usual assumptions of DIHB analysis are not reliable for the concrete-like specimen. An inverse approach is nevertheless possible to give more accurate results.