This paper presents a methodology for the automatic detection of sub-optimal performance in UMTS networks based on drive-test measurements. The proposed methodology can be seen as a first step towards the more ambitious objective of implementing Self-Organising Network (SON) concepts in future wireless networks. In fact, the lessons learnt from real UMTS case studies enable the identification of key issues related to self-optimisation whose relevance can be more difficult to attain theoretically. In particular, this paper presents algorithms targeting the identification of coverage holes, cell overlaps and cell overshooting. Relevant quality indicators are identified in each case. A case study with real data extracted from a medium-size European city is presented to illustrate the methodology.