Covariate balance is crucial for causal comparisons in observational studies. Recently we provided a unified framework -- the balancing weights-- to balance covariates between treatment groups via propensity score weighting. These weights incorporate the propensity score to weight each group to an analyst-selected target population, and include several commonly used weighting schemes such as inverse-probability weight and trimming as special cases. Within this class, we advocate the overlap weights, which weigh each unit by its probability of being assigned to the opposite group. The overlap weights focus on a target population that is close to clinical equipoise, and possess attractive theoretical and operational properties, such as optimal asymptotic variance and exact balance. In this talk, we will discuss several important aspects and extension of the overlap weights in theory and practice, including variance estimation, ratio estimands, multiple treatments and subgroup analysis. All discussions will be in the context of real applications.