The effects of precipitates in the microstructure of an AA7075 aluminium alloy on: the dislocation behaviour, dislocation structure and dislocation structure evolution during plastic deformation were investigated using the physical yield criterion model by van Liempt and Sietsma. The model was in fact applied for the first time to aluminium and an AA7075 aluminium alloy. By constructing an extended Kocks-Mecking plot from the measured tensile and interrupted tensile data the dislocation density, average dislocation segment length and physical yield stress were determined. The aim of the present study was to get a better understanding of the role of precipitates in the evolution of the dislocation structure during plastic deformation. The recovery of anelastic strain during loading and subsequent unloading after plastic deformation was investigated as well. The yield criterion was used to study the anelastic loading, while a constitutive unloading model by Torkabadi et al. was used to study the unloading behaviour. The models were combined to define the fraction of unrecoverable anelastic strain. The anelastic strain is related to springback. Therefore a better understanding of the anelastic strain could be useful for making better predictions of springback after metal forming. The physical yield criterion was further extended by incorporating a continuous uniform dislocation segment length distribution. Insight on the dislocation segment length distribution, and the evolution thereof during plastic deformation, could help to better understand the mechanical behaviour of metals. The evolution of the dislocation structure during plastic deformation is impeded by the presence of precipitates in the microstructure. Therefore, the physical interpretation of α in the Taylor equation which quantifies the dislocation structure, proposed by Arachebelata et al., was modified to incorporate the effects of precipitates in the microstructure. The dislocation structure parameter α should remain constant. However, the average dislocation segment length obtained from the extended Kocks-Mecking plot does not decrease sufficiently to accommodate a constant α. The introduction of the length between precipitates, which is independent from work hardening, into the Taylor equation ensures that α does remain constant, whilst in addition it provides an estimate of the distance between the precipitates. Anelastic strain, caused by reversible glide of dislocations in the pre-yield regime, is introduced into or recovered from the metal during loading and unloading respectively. The anelastic unloading strain determined from the interrupted tensile tests was found to be smaller than what was expected according to the model. Three possible causes were identified: not all Frank-Read sources are at their critical state when unloading is initiated, dislocation loops propagating through the crystal undo portions of the anelastic unloading strain and the retracting dislocations remain stuck behind obstacles whilst retracting under the reducing applied stress. The constitutive model was found to be not suitable for studying the dislocation behaviour during unloading because the change of dislocation segment length with plastic deformation is not accounted for. The development of a physical unloading model is therefore recommended. The distribution of the dislocation segment lengths could explain the non-zero value of the work hardening rate Θ at the abrupt change of slope between the pre and post-yield regime in the extended Kocks-Mecking plot. The value of Θ at this point could be an indication of the distribution width. The continuous uniform distribution is a rather unrealistic description of the dislocation segment length distribution. Therefore, other distribution types are proposed for the further development of the dislocation segment length distribution model of the physical yield criterion.