THE EXISTENCE OF CALLABLE bonds has been somewhat of a puzzle. The call feature provides firms with a valuable interest rate option, but the value of this option should be exactly offset by the creditor's demand for a higher interest rate as compensation. One explanation for the call provision from Myers [4], Bodie and Taggart [3], and Barnea, Haugen, and Senbet [2] is that callable bonds may be used to reduce some agency costs. Another explanation from Robbins and Schatzberg [5] is that callable bonds can signal a firm's better prospects in the presence of asymmetric information. One possible problem with the agency cost and signalling explanations of callable bonds is that short-term debt may serve the same functions in all four studies. Indeed, short-term debt is equivalent to long-term debt in the theoretical sections of the agency cost studies. However, Robbins and Schatzberg (hereafter RS) suggest that short-term debt fails to provide the risk-sharing opportunities inherent in callable bonds. Thus, they suggest that a signalling explanation provides a unique role for callable bonds. One limitation of RS's results is that they are based on an example that makes a series of strong assumptions about a two-stage lottery faced by and firms. In order for their model to have general applicability, it would be desirable to demonstrate that their results hold under more general assumptions. The results of this study suggest that their findings cannot be generalized without further assumptions. Callable bonds do not yield a separating equilibrium result when only a minor change is made in two of the lottery probabilities, suggesting that their results are far from robust to minor changes in assumptions. Moreover, callable bonds may fail to generate a separating equilibrium even for some good-news firms that conform to RS's original model because bad-news firms that do not meet their assumptions may attempt to mimic good-news firms that do. The risk-sharing result is also very sensitive to the specific assumptions made by Robbins and Schatzberg; specifically, callable bonds increase the risk borne by management if one change is made in the numeric values of one of the lottery payouts. The paper begins with a review of the relevant portions of Robbins and Schatzberg's model, Section II provides a counterexample in which callable bonds fail to provide a separating equilibrium, Section III provides a counterexample in which callable bonds fail to provide risk-sharing debt, and the paper concludes with a summary.