This paper introduces Coarsened Exact Matching (CEM), a new method for causal inference that avoids the need for balance checking. CEM is derived from the Monotonic Imbalance Bounding (MIB) class of matching methods. Unlike traditional matching methods, CEM guarantees that the imbalance between treated and control groups will not exceed a pre-specified level, which is chosen based on substantive knowledge. This ensures that the method is robust to model dependence and statistical bias, while being easy to use and understand. CEM works by first coarsening each variable into intervals, then performing exact matching on the coarsened data. The coarsened data are then discarded, and the original values of the matched data are retained. CEM has several desirable statistical properties, including the ability to handle a wide range of data types and to reduce model dependence and statistical bias. The paper also discusses various extensions of CEM and provides an empirical illustration of its effectiveness. The authors also make available open-source software for R, Stata, and SPSS that implements all their suggestions. CEM is particularly useful in observational studies where the treatment assignment mechanism is unknown or ambiguous. The method is also robust to measurement error and can be used to restrict data to common empirical support. Overall, CEM is a powerful and flexible method for causal inference that avoids the need for balance checking and is particularly useful in situations where traditional matching methods fail.This paper introduces Coarsened Exact Matching (CEM), a new method for causal inference that avoids the need for balance checking. CEM is derived from the Monotonic Imbalance Bounding (MIB) class of matching methods. Unlike traditional matching methods, CEM guarantees that the imbalance between treated and control groups will not exceed a pre-specified level, which is chosen based on substantive knowledge. This ensures that the method is robust to model dependence and statistical bias, while being easy to use and understand. CEM works by first coarsening each variable into intervals, then performing exact matching on the coarsened data. The coarsened data are then discarded, and the original values of the matched data are retained. CEM has several desirable statistical properties, including the ability to handle a wide range of data types and to reduce model dependence and statistical bias. The paper also discusses various extensions of CEM and provides an empirical illustration of its effectiveness. The authors also make available open-source software for R, Stata, and SPSS that implements all their suggestions. CEM is particularly useful in observational studies where the treatment assignment mechanism is unknown or ambiguous. The method is also robust to measurement error and can be used to restrict data to common empirical support. Overall, CEM is a powerful and flexible method for causal inference that avoids the need for balance checking and is particularly useful in situations where traditional matching methods fail.