Recent studies by Erickson and Whited (2000) and Cummins, Hassett, and Oliner (2006) claim that positive estimates of investment-cash flow sensitivity reported in earlier papers are mere artifacts of measurement error in Tobin's q. This paper presents evidence refuting such claims. The Cummins, Hassett, and Oliner (2006) findings suffer from three problems: 1. Several implementational shortcomings, viz., incorrect choice of analyst expectations windows, an incorrect valuation formula, and use of earnings instead of cash flows while valuing equity. 2. Unnecessarily restricting lag lengths of candidate instruments. 3. An apparent data discrepancy in the relative timing of the investment and cash flow series. When these problems are accounted for, their findings of the superiority of an analyst forecast-based measure of q and the insignificance of cash flow effects cease to hold. Erickson and Whited's (2000) findings based on using higher order moments are also contradicted at three levels: 1. Their reported findings are materially sensitive to small changes in variable constructions. Specifically, using variable definitions that are standard in the literature yields contrary results. 2. Extending the sample from four years (1992-1995) to twenty-two (1982-2003) leads to specification test failures and consistently positive estimates of investment-cash flow sensitivity. 3. Extending their GMM3 estimator to incorporate the information in analyst expectations also yields contradictory results. The availability of two mismeasured proxies for q - one based on stock market prices and the other on analyst expectations - allows the use of lags of one as instruments for the other. The dynamic panel data GMM approach of Arellano and Bond (1991, 1998) yields models that are at once powerful and well-specified, and estimated investment-cash flow sensitivity is consistently positive and significant. The evidence also confirms the earlier findings of Allayannis and Mozumdar (2004) and Bhagat Moyen, and Suh (2005) that negative cash flow observations are materially different from the rest of the sample. Sweeping generalizations about the investment-cash flow effect being a myth therefore appear unwarranted.