This paper addresses the issue of estimating dynamic treatment effects in event studies, where treatment effects can vary across units due to differences in treatment timing. The authors show that the coefficients on leads and lags in two-way fixed effects regressions can be contaminated by effects from other periods, leading to apparent pretrends that may not accurately reflect the true treatment effects. They propose an alternative estimator that is free from this contamination and demonstrate its effectiveness through an empirical application. The paper also provides a decomposition of the relative period coefficients to illustrate how treatment effects from other periods can contaminate the estimates. The authors further develop a regression-based method that uses cohort weights to estimate dynamic treatment effects, which is more robust to treatment effect heterogeneity. The method is illustrated using data from the Health and Retirement Study to estimate the dynamic effects of hospitalization on out-of-pocket medical spending and labor earnings. The results show that the proposed method yields similar findings to those obtained using two-way fixed effects regressions but avoids the issues of contamination and provides more interpretable estimates.This paper addresses the issue of estimating dynamic treatment effects in event studies, where treatment effects can vary across units due to differences in treatment timing. The authors show that the coefficients on leads and lags in two-way fixed effects regressions can be contaminated by effects from other periods, leading to apparent pretrends that may not accurately reflect the true treatment effects. They propose an alternative estimator that is free from this contamination and demonstrate its effectiveness through an empirical application. The paper also provides a decomposition of the relative period coefficients to illustrate how treatment effects from other periods can contaminate the estimates. The authors further develop a regression-based method that uses cohort weights to estimate dynamic treatment effects, which is more robust to treatment effect heterogeneity. The method is illustrated using data from the Health and Retirement Study to estimate the dynamic effects of hospitalization on out-of-pocket medical spending and labor earnings. The results show that the proposed method yields similar findings to those obtained using two-way fixed effects regressions but avoids the issues of contamination and provides more interpretable estimates.