Permanent deformation is one of the failure modes considered in the analysis and design of flexible pavements. ERAPave performance prediction (PP) which is a mechanistic empirical (ME) pavement design tool utilises two distinct models for the prediction of permanent deformation in the bound and unbound granular layers including subgrade. This paper aims to calibrate these models using pavement response and performance data from accelerated pavement testing (APT) structures. Material properties such as layer modulus were established through an optimisation that involves both falling weight deflectometer (FWD) and pavement response measurements. Based on the predicted performance results, a separate set of calibration was performed for permanent deformation development in moist and wet moisture conditions. The calibrated models have resulted in predictions that are in good agreement with observed performances. Furthermore, the model parameters successfully captured the initial densification behaviour and the associated sensitivity with axle load level.