Remarkably, there has been no analytic consideration of the origin of the widely accepted limitation on pulsed thermography defect detection performance that defect aspect ratio (diameter/depth) must exceed two. A simple modification of the one-dimensional expression for the contrast of a layer has been developed that provides an accurate prediction of peak contrast and contrast peak time for defects of all aspect ratios. The modification is based on the assumption that defect image contrast is limited by the diffusion of trapped heat from above the defect to the defect edge. The analytic model results have been shown to agree with numerical modelling for a range of materials, including composites which have anisotropic thermal conductivity. Both peak defect image contrast and occurrence time are found to be systematic functions of defect aspect ratio but, in addition, contrast falls with defect depth. The rule-of -thumb is only true for a limited range of defect depths and it depends critically on absorbed excitation/flash energy. The analytic impulse excitation model can be developed for other forms of excitation (long pulse or periodic) or used to model time derivative processing of thermography data. The possibility of developing a thermographic NDE expert system will be discussed.