The cracked chevron notched Brazilian disc (CCNBD) specimen, suggested by the International Society for Rock Mechanics for testing mode I fracture toughness of rocks, usually yields rather conservative toughness measurements, and the reasons have not been fully explored. In this study, the CCNBD method is compared with the cracked chevron notched semicircular bending (CCNSCB) method in the fracture process zone (FPZ) and its influence on the fracture toughness measurement. Theoretical analysis reveals that the FPZ is longer in the CCNBD specimen than in the CCNSCB specimen using a relatively large support span, the toughness measurement using the former is affected more seriously by the presence of FPZ, and thus the CCNBD method is usually, more or less, conservative compared with the CCNSCB method. These inferences are further validated by experimental results, which indicate that the CCNBD test indeed produces much lower fracture toughness values and even the results of 75‐mm radius CCNBD specimens are still lower than those of 25‐mm radius CCNSCB specimens. Consequently, due to smaller FPZ, the CCNSCB specimen with a relatively large span is more likely to produce comparably accurate or representative toughness value, and it may be more suitable than the CCNBD specimen for the engineering applications that require more representative or less conservative fracture toughness.