The paper proposes robust methods for estimating and inferring the effect of a treatment variable on a scalar outcome when there are many controls. The authors develop a novel estimation and inference method called the "post-double-selection" method, which involves two stages of variable selection followed by a final estimation step. The first stage selects control variables that predict the treatment, and the second stage selects additional variables that predict the outcome. This approach allows for imperfect selection of controls and provides uniformly valid confidence intervals across a wide class of models. The method is applicable to Lasso-type methods and other model selection techniques that can find a sparse model with good approximation properties. The paper includes theoretical results showing consistency and asymptotic normality of the estimator, as well as numerical simulations and an application to the effect of abortion on crime rates. The main contributions include a robust estimation and inference method for high-dimensional controls, and the development of supporting theory.The paper proposes robust methods for estimating and inferring the effect of a treatment variable on a scalar outcome when there are many controls. The authors develop a novel estimation and inference method called the "post-double-selection" method, which involves two stages of variable selection followed by a final estimation step. The first stage selects control variables that predict the treatment, and the second stage selects additional variables that predict the outcome. This approach allows for imperfect selection of controls and provides uniformly valid confidence intervals across a wide class of models. The method is applicable to Lasso-type methods and other model selection techniques that can find a sparse model with good approximation properties. The paper includes theoretical results showing consistency and asymptotic normality of the estimator, as well as numerical simulations and an application to the effect of abortion on crime rates. The main contributions include a robust estimation and inference method for high-dimensional controls, and the development of supporting theory.