摘要：We focus on the construction of con_dence interval for several parameters: a single proportion from a binomial distribution, an unknown population size and an unknown population total in hypergeometric distribution and the odds ratio. These parameters are widely used in practice including the binomial experiment, the survey sampling and the capture-recapture sampling and the case-control study. The common feature is that they all have a _nite sample space. The currently used intervals either have an incorrect (low) coverage probability or are conservative or are di_cult to compute. For each parameter, we describe an exact admissible two-sided interval and discuss the implementation in R.