Modelling crack growth to determine inspection periods and residual life demands consideration of at least three load interaction phenomena - fatigue crack closure, near-tip residual stress and crack-tip blunting/re-sharpening. The first may be modelled as essentially a wake-sensitive phenomenon whereby an increase in closure always demands to crack extension (for wake development), while decrease or elimination can be instantaneously induced by blunting. The effect of near-tip residual stress has for long been wrongly attributed to the monotonic plastic zone. We can be certain now, that the residual stress effect manifests through moderation of crack-tip diffusion kinetics of active species such as hydrogen at room temperature and oxidants at high temperature. Being a surface phenomenon, the effect is sensitive to cyclic not monotonic) plastic zone response and is, therefore, a 'cycle-sequenced' dependent phenomenon (it does not require crack extension). More importantly, for the same reason, its effect progressively increases with decreasing crack growth rate, thereby effectively controlling threshold stress intensity (and therefore sub-Paris Regime growth rate) in atmospheric fatigue. A method was developed to model near-tip residual stress as a function of load history. An experimental procedure was developed to determine threshold stress intensity as a function of this stress value. The two appear to show an excellent correlation. Experiments to demonstrate cycle-sequenced sensitivity of threshold stress intensity and also the disappearance and fast recovery of closure from crack-tip blunting appear to be supported by fractographic evidence. A good test of a crack growth model is a check of its 'microscopic consistency', whereby, the model is checked for its ability to correctly estimate the contribution to the actual crack extension of cycles of different magnitude in a load spectrum. Such a test was devised by programmed adaptations of the TWIST load spectrum. The results of these experiments are encouraging in that they appear to support the new modelling approach as shown by quantitative fractography of fractures obtained under two extremes in the arrangement of the TWIST spectrum. The study appears to show promise of the new approach to handling spectrum load crack growth in atmospheric fatigue. It is leading to a new approach to testing as well as analyses.