AbstractConventionally, the fatigue threshold of a long‐crack is obtained by load shedding using a constant normalizedK‐gradient, as recommended by ASTM E‐647. However, this load shedding procedure often causes the crack opening displacement to decrease with increasing crack length, which may trigger crack closure. In this study, load shedding tests were conducted in load control following a power‐law load shedding schedule such that the crack opening displacement was kept at a relatively constant level. Using this new testing procedure, it is shown that crack closure is not always as high as that associated with the ASTM recommended procedure at a given ΔK.Comparisons of fatigue crack growth rates under identical testing parameters, but with different closure levels produced by the two load shedding procedures, have been made for several structural alloys. The extrinsic shielding of the crack tip zone via crack closure has also been examined using an energy approach for these alloys. On the basis of these analyses, the true effective stress intensity range is evaluated for fatigue crack propagation and the role of crack closure in creating a fatigue threshold is re