In this paper, we examine stationarity properties of data on budget deficits for a cluster of twenty-seven developing countries (D-27) for the period spanning 1970 to 2003. It has been argued in the literature that this statistical property correlates well with the economic property of sustainability of the budget deficit. The univariate unit root tests indicated a non-stationary process of I(1) with the exception of three countries. However, the non-stationary properties were rejected when the panel unit roots procedures were adopted. Since panel procedures provide greater power, the statistical evidence favors stationarity. This in turn suggests that budgets deficits in our sample of countries are sustainable, an important conclusion with many real world economic implications. The conflict between single country results and panel results suggests that univariate procedures may lead to the wrong conclusions.