Short-rod fracture toughness specimens of various sizes were used in toughness measurements on six different metals. The toughness values were observed to be independent of the specimen size, within experimental scatter, whenever the specimen diameter exceeded 1.25(KIcSR/σys)2. The test records were analyzed by a linear elastic fracture mechanics (LEFM) method and also by an elastic-plastic method. When the specimen size was too small to yield a size-independent result, the LEFM analysis always tended toward conservative (smaller) toughness values, while the elastic-plastic analysis always tended toward nonconservative values. Thus, the two analysis techniques apparently establish upper and lower bounds for the toughness when the specimen is too small for a valid test. It was also found that the slots which form the chevron in the specimen must either be very thin or have sharply pointed slot bottoms (≤60° included angle) to maintain good plane-strain constraint along the crack front.