This paper presents an empirical test of the Brennan Kraus (1987) hypothesis of convertible bond financing, according to which firms signal their volatility by their choice of terms of the convertible issue. With additional assumptions about the nature of investors’ prior beliefs about firm types the model predicts that the announcement period return will be positive related to the face value of the convertible, and negatively related to the fraction of the firm accruing to the convertible holders on conversion. The empirical evidence for a sample of public issues strongly supports these predictions, while that for a sample of private placements does not, which is consistent with problems of information asymmetry being important for the former but not for the latter.