Abstract This study investigates the performance of firms with extremely high levels of market to sales value (“concept stocks”). To many observers, these stocks appear overvalued. However, proponents argue that because of their unique characteristics, traditional pricing models fail to value these firms correctly. Ex post, the debate can be resolved through an analysis of the long-term performance of concept stocks. En route to testing the implied overpricing hypothesis we document several important findings. First, the identity and characteristics of concept stocks have changed markedly over time. Although the obvious recent examples are internet and biotech stocks, concept stocks vary widely by industry over the past four decades. The industries containing the most popular concept stocks evolve from oil and gas extraction in the 1960s and 1970s, to computer and office equipment in the 1980s, and to computer-related services in the 1990s. Second, although concept stocks tend to be young, small, growth stocks in the 1990s, they exhibit a wide range of characteristics throughout the sample period. Third, the relative pricing of concept stocks (compared to either a control sample or the entire population) has changed dramatically over time. The average concept stock sold for approximately three times sales in the late 1960s and 1970s, five times sales in the 1980s and nearly 17 times sales in the 1990s. Finally, we find evidence supporting the overpricing hypothesis. Concept stocks under-perform significantly in the long run. This under-performance is more severe for Nasdaq firms and in the most recent two decades. The results are separate from glamour, IPO, industry, or contrarian effects and remain after an extensive sensitivity analysis.