In aerospace industry and other major mechanical industry systems, engineering components that are subjected to cyclic loads often lead to progressive crack growth that eventually results in struc-tural fracture. The damage tolerance design which is based on the assumption of pre-existed flaws in a structure is an important approach in aircraft industry since it is impossible to have flaw-free manufactured components.In this thesis work, an evaluation of crack propagation models was carried out. Fatigue crack growth threshold and fatigue crack growth rate models were evaluated. A method to present ex-perimental data available was developed to evaluate uncertainties in fatigue life models for more accurate predictions. Currently, a software that is used for predicting crack propagation life is NASGRO. The study has been made for two types of materials: a nickel-iron-based alloy (Inconel 718 forging) and titanium alloys (Ti 6-4 both forging and casting).A threshold model is in the normal case developed for each temperature. A method to model fatigue threshold (ΔKth) has been suggested by assuming temperature independence of ΔKth. In this method, a new threshold model was created by making use of an A/P (Actual/Predicted) plot so that all measured threshold values are on the conservative side of the minimum model. With this method, an understanding of fatigue threshold model was improved over the other method due to the possibilities to model ΔKth with average and minimum threshold values for each load ratio (ΔKth, R).Moreover, a method to investigate which set of parameters that best represent the crack growth behaviour has been suggested. In this method the best set of parameters were chosen to be the set of parameters giving the best fit to the available (da/dN, ΔK) points. The comparison between this method and the method with the set of parameters that give minimum scatter in the A/P values was done.Crack growth rate da/dN log curves were plotted as function of stress intensity range ΔK for R-ratio values ranging from -2 to 0.9 for the two different methods. A distinctive difference between the two methods was observed in Paris region at high temperatures (5500C-6500C) which becomes more obvious at lower R-ratios. Predicting crack propagation rate model with set of parameters giving minimum standard deviation in da/dN points was shown to be less conservative than that of parameter sets giving lowest scatter in A/P. Using both evaluation methods, da/dN versus ΔK plots of Inconel718forging were compared to da/dN (ΔK) plots for the pre-existing data at 5500C for R-ratios ranging from 0 to 0.8. An overall R-ratio influence was observed throughout for both ΔKth and da/dN.