At a recent conference on vocabulary measurement and testing (held at the University of the West of England, Bristol, in January 2004) those who took part were struck by the way independent researchers were constantly returning to the same theme in their work; that of the validity of the measures we were using. This trend was not restricted to new efforts to model and measure vocabulary knowledge, as in attempts to characterise vocabulary depth. But they extended too to areas such as vocabulary breadth where, at first sight, the measures we have appear extremely well founded and reliable, and where the qualities of the tests appear to be well known. This chapter, then, looks at a range of factors that have been shown to affect the validity of vocabulary measures and it provides an oversight of the issues which will be addressed again and again in subsequent chapters. These issues include the selection of vocabulary items, the attitude of the subjects involved, what is counted as a word, and various aspects of what it means to know a word (multiple measures, the use of the first language, and vocabulary in use). There are satisfactory ways of dealing with these factors and they mostly relate to having a clear idea of the purpose of the measure.