In this article, we consider the problem of estimating some lifetime parameters based on adaptive progressive Type-II censored sample from the inverse Weibull distribution. Maximum likelihood (ML) and Bayesian approaches are used to estimate the unknown parameters, coefficient of variation, reliability and hazard functions. The Bayes estimators are obtained using both symmetric and asymmetric loss functions. However, the Bayes estimators do not exist in an explicit form, Markov Chain Monte Carlo (MCMC) method is used to generate samples from the posterior distribution. Gibbs sampling within Metropolis. Hastings is applied to estimate the lifetime parameters. Furthermore, asymptotic normality of the ML and MCMC method are employed to construct the corresponding confidence intervals. The delta method is used to estimate the variances of coefficient of variation, reliability and hazard functions. Further, Bayesian two-sample prediction of the future order statistics as well as the future lower record values are discussed. Proposed methods of estimation and prediction are compared using Monte Carlo simulation study. Finally a real data set is analyzed for illustration purposes.