This study addresses issues that arise when using daily NASDAQ security returns in event studies. Its objective is to provide benchmarks for assessing the impact of alternative research design choices. The results indicate that test statistics commonly used in event studies are subject to varying degrees of misspecification in samples of NASDAO securities. Misspecification is most severe for traditional z-statistics based on standardized abnormal returns. Across a wide variety of event study settings the rank statistic introduced in Corrado (1989) is consistently well-specified and quite powerful in detecting abnormal performance.