In altitude test facility (ATF) operation, the requirements to control humidity to generate defined icing conditions are gaining more and more importance. In this context, the ability to predict humidity and condensation becomes a fundamental part of ATF control. For this purpose, classical nucleation theory has been applied in combination with in situ measurements to derive a model suitable to predict the onset of condensation during very low temperature ATF operation. The model parameters have been acquired inside the ATF of the University of Stuttgart downstream of its air coolers. This makes the application or assumption of generalized atmospheric aerosol data unnecessary. Polydisperse nano aerosol distributions were measured and statistically evaluated, showing that a constant distribution of nano aerosol particle size can be assumed. The composition of the ingested nanoparticles was analyzed and Arizona test dust was chosen as a valid substitute material for the application in the prediction model leading to a conservative prediction. The approach has been successfully verified using optical measurements during ATF testing. Its prediction accuracy fulfills the requirements of ATF control for a variety of icing conditions in component and engine altitude testing.