The value of inter-comparisons can be limited by the difficulty of auditing model differences back to the processes that caused them. Conditional analysis methods can be used to improve this auditability. Another recent improvement is the development of a ‘hybrid’ model that facilitates more rapid inter-comparisons between ADMS and AERMOD. An example of using conditional analysis to compare models is given for a tall stack with receptors in the near- and far-field. Differences between the models are related to different predictions of boundary-layer height; these predictions are compared using a ‘Dispersion Calendar’ that shows how height differences vary for different dispersion conditions.