Traditional wireless location verification (authentication) is only feasible under the assumption that radio propagation is described by simple time-independent mathematical models. A similar situation applies to location acquisition, albeit to a lesser extent. However, in real-world situations, channel conditions are rarely well-described by simple mathematical models. In this thesis, novel location verification and acquisition techniques that integrate machine learning algorithms into the decision process are designed, analysed, and tested. Through the use of both simulated and experimental data, it is shown how the novel solutions developed remain operational in unknown time-varying channel conditions, thus making them superior to existing solutions, and more importantly, deployable in real-world scenarios. Location verification will be of growing importance for a host of emerging wireless applications in which location information plays a pivotal role. The location verification solutions offered in this thesis are the first to be tested against experimental data and the first to invoke machine learning algorithms. As such, they likely form the foundation for all future verification algorithms.