Using approximately 10,000 firms jointly covered by Compustat and CRSP from 1974–1993, we find substantial differences in the SIC codes designated by the two databases. More than 36 percent of the classifications disagree at the two-digit level and nearly 80 percent disagree at the four-digit level. We examine the impact of these differences upon financial research in several ways. First, we show that the classification of utilities, financial firms, and conglomerate acquisitions are affected by the choice of CRSP vs. Compustat SIC codes. Second, we show that industry classification matters in financial research by illustrating that size- and industry-matched comparisons are more powerful than pure size matches. Third, we test the specification and power of Compustat vs. CRSP classifications by simulating a typical financial experiment in which sample firms are matched to control firms by industry. We find that: i) Compustat matched samples are more powerful than CRSP matched samples in detecting abnormal performance; ii) nonparametric tests outperform parametric tests; and iii) four-digit SIC code matches are more powerful than two-digit SIC code matches. These results are robust to the inclusion or exclusion of extreme values, and hold for both NYSE/AMEX and Nasdaq firms.