The objective of this study was to obtain a better understanding of the delayed hydride cracking (DHC) of Zr-2.5Nb alloy. The DHC model has some defects: first, it cannot explain why the DHC velocity (DHCV) becomes constant regardless of an applied stress intensity factor, even though the stress gradient is affected by the applied stress intensity factor at the notch tip. Second, it cannot explain why the DHCV has a strong dependence on the method of approaching the test temperature by a cool-down or a heating-up, even under the same stress gradient, and third, it cannot predict any hydride size effect on the DHC velocity. The DHC tests were conducted on Zr-2.5Nb compact tension specimens with the test temperatures reached by a heating-up method and a cool-down method. Crack velocities were measured in hydrided specimens, which were cooled from solution-treatment temperatures at different rates by being furnace-cooled, water-quenched, and liquid nitrogen-quenched. The resulting hydride size, morphology, and distributions were examined by optical metallography. It was found that fast cooling rates, which produce very finely dispersed hydrides, result in higher crack growth rates. This different DHC behavior of the Zr-2.5Nb tube with the cooling rate after a homogenization treatment is due to the precipitation of the $gamma$-hydrides only in the water-quenched Zr-2.5Nb tube. This experiment will provide supporting evidence that the terminal solid solubility of a dissolution (TSSD) of $gamma$-hydrides is higher than that of $delta$-hydrides.