A number of researchers have recently proposed a variety of different `vulnerability'measures designed to capture the welfare consequences of risk for poor households, and also proposed a variety of different approaches to estimating these various measures of household vulnerability. However, it's possible to `mix-and-match'estimators and measures. Here we conduct Monte Carlo experiments designed to explore the performance of different estimators with different measures, under different assumptions regarding the underlying economic environment. We find that when the environment is stationary, and consumption expenditures are measured without error, that the best estimator is one proposed by Chaudhuri (2001), regardless of what measure of vulnerability is employed. If the vulnerability measure is risk-sensitive, but consumption is measured with error, a simple estimator proposed by the authors(2003) generally performs best. However, when the distribution of consumption is non-stationary, a modification of an estimator proposed by Pritchett et al. (2000) performs best. Future research should focus on combining the efficiency of the Chauduri estimator with the good properties of the authors (in environments with measurement error), and Pritchett (in non-stationary environments) estimators. However, even with present technology estimating vulnerability is simple, and much more informative, and useful than are static poverty measures, provided one has at least two rounds of panel data.