AbstractCreep‐fatigue crack growth (C‐FCG) rates in a P91 steel at 625°C were correlated as the average time rate of crack growth during hold time, (da/dt)avg, with (Ct)avg. At 60‐second hold time, the rates were lower than for 600‐second hold time. At 600‐second hold time, the crack growth rates converged on to the creep crack growth rate (CCGR) trend. Thus, the CCGR trend represents the upper bound for time‐dependent crack growth rates in P91 materials. The analytical expressions based on considering just the elastic and secondary creep deformation rates overestimated the magnitudes of (Ct)avg by as much as a factor of 10 for the 600‐second hold time tests. After accounting for the effects of cyclic plasticity during unloading, and accounting for only partial reversal of creep strains accumulated during hold time, the estimates of (Ct)avg compared well with the measured values. CR represents the extent of crack tip creep strain reversal, and tpl is the time required for the crack tip creep zone during the hold time to become equivalent in size to the cyclic plastic zone in terms of stress carried by that region. Together, these parameters accurately account for the effects of crack tip cyclic plasticity on the magnitude of (Ct)avg.Both tpl and CR depend on material properties, and the latter also depends on the hold time. A parameter ϕ is introduced that is dependent only on material properties and from which CR can be estimated for a given hold time. tpl and ϕ can be reported as part of the test results from C‐FCG testing.