This article proposes a robust test for weak instruments in linear instrumental variables (IV) regression that is robust to heteroscedasticity, autocorrelation, and clustering. The test statistic is a scaled nonrobust first-stage F statistic. Instruments are considered weak when the two-stage least squares (TSLS) or limited information maximum likelihood (LIML) Nagar bias is large relative to a benchmark. The test is applied to estimate the elasticity of intertemporal substitution (EIS), where it cannot reject the null of weak instruments in a larger number of countries than the test proposed by Stock and Yogo (2005). The test is robust to heteroscedasticity, serial correlation, and clustering, and is implemented using generalized and simplified procedures. The generalized procedure is more powerful but computationally more demanding, while the simplified procedure is conservative. The test is applied to an empirical example, the IV estimation of the EIS, and is shown to be effective in identifying weak instruments in the presence of heteroscedasticity and serial correlation. The results confirm Yogo's (2004) finding that the EIS is small and close to zero. The test is robust to heteroscedasticity, serial correlation, and clustering, and is implemented using a scaled version of the regular F statistic. Critical values depend on the covariance matrix of the reduced form coefficients and errors. The test is applied to a variety of models and is shown to be effective in identifying weak instruments in the presence of heteroscedasticity and serial correlation. The results confirm Yogo's (2004) finding that the EIS is small and close to zero.This article proposes a robust test for weak instruments in linear instrumental variables (IV) regression that is robust to heteroscedasticity, autocorrelation, and clustering. The test statistic is a scaled nonrobust first-stage F statistic. Instruments are considered weak when the two-stage least squares (TSLS) or limited information maximum likelihood (LIML) Nagar bias is large relative to a benchmark. The test is applied to estimate the elasticity of intertemporal substitution (EIS), where it cannot reject the null of weak instruments in a larger number of countries than the test proposed by Stock and Yogo (2005). The test is robust to heteroscedasticity, serial correlation, and clustering, and is implemented using generalized and simplified procedures. The generalized procedure is more powerful but computationally more demanding, while the simplified procedure is conservative. The test is applied to an empirical example, the IV estimation of the EIS, and is shown to be effective in identifying weak instruments in the presence of heteroscedasticity and serial correlation. The results confirm Yogo's (2004) finding that the EIS is small and close to zero. The test is robust to heteroscedasticity, serial correlation, and clustering, and is implemented using a scaled version of the regular F statistic. Critical values depend on the covariance matrix of the reduced form coefficients and errors. The test is applied to a variety of models and is shown to be effective in identifying weak instruments in the presence of heteroscedasticity and serial correlation. The results confirm Yogo's (2004) finding that the EIS is small and close to zero.