This paper mainly focuses on the approach taken at the Bank of Greece regarding the application of stochastic methods to debt sustainability analysis, providing also a discussion of alternative options. Caution is advised in the way that stochastic methods are made operational, as they are far from exact and rely on assumptions of various degrees of plausibility, which are often not stated explicitly. A Monte Carlo exercise reveals that under the approach taken by the European Commission, the measurement of dispersion can be subject to significant bias, ranging from an over-estimation by 45% to an under-estimation in excess of 80%, depending on the time-series properties of the data.