Some aspects of crack closure are reviewed and examined with respect to one's ability to predict fatigue crack propagation rates in engineering alloys. The development of artificial crack closure with the placement of shim stock in the crack mouth provided nonconservative estimates of crack growth rates as compared with those rates for specimens influenced by natural closure processes. A reevaluation of the ΔK-decreasing (constant stress rati o) test procedures suggests that this method may introduce abnormally high closure levels and generate nonconservative values of ΔKth. The implication of comparing such long crack data with results from short crack samples is discussed.