A simple method is described for the estimation of predissociation broadening in the infrared spectra of hydrogen-bonded species, which avoids the complicated mathematics of the original treatment by Coulson and Robertson. The two treatments give equivalent results in a certain limiting case, and this provides a check on the correctness of each. Numerical values are calculated for the excited vibrational level widths of Me2O. HCl, which although much greater than were found previously, provide no evidence that predissociation broadening should be observable experimentally. However, recent work on combination bands forces the conclusion that in practice the predissociation broadening must be greater than that calculated here and, a fortiori, greater than the original estimate by Coulson and Robertson. It still seems unlikely that Stepanov's predissociation effect should be observable, but this conclusion is not absolutely secure, in the absence of reliable information about the nature of the coupling between ν(Cl—H) and ν(Cl—H … O) stretching modes of vibration. The major criticism of the present simplified treatment of vibrational predissociation applies with equal force to other attempts to calculate by perturbation theory the lifetimes of unstable molecules.