Instrumented impact tests were carried out on an API X-52 steel and its weldment by using Charpy V-notch (CVN) specimens and pre-cracked Charpy V-notch (PCVN) specimens. The physical meaning of the energy absorbed up to the maximum load in the load-deflection curve, E i, and several factors affecting the E i value were studied. It was found that for the CVN specimen the E i value expressed the crack initiation energy, and for the PCVN specimen the E i value was the sum of the crack initiation energy and part of energy of crack propagation. The E i value decreased with decreasing test temperature no matter whether the material was homogeneous or heterogeneous. For the CVN specimen, there were three characteristic temperatures in the E i- T°C relation curve: (1) T b—the temperature below which cleavage crack initiation would occur; (2) T d—the temperature above which a ductile crack initiated; (3) T r—the higher limit of temperature of the middle shelf in the E i- T°C relation curve. However, there were only two characteristic temperatures, T b and T d, for the PCVN specimen. The E i value was affected by specimen orientation, chemical composition of weld metal, and constraint. For ductile fracture, E i E t and E p E t values ( E t = E i + E p, E p—post-maximum load energy in the load-deflection curve) were insensitive to microstructure and material state, such as rolling or casting, over a certain range of material strength. The E i E t value decreased with increasing constraint at the same time as the E E p E t value increased. A peak of the E i E t values appeared at the T d temperature in the E i- T°C relation curve, whilst, on the other hand, a valley point of the E p E t values then occurred.