It uses a first-order Taylor expansion to approximate the variance of a function of an asymptotically normal estimator.

It is a method to generate bootstrap samples.

It is used for non-parametric density estimation.

It requires the exact distribution of $\hat{\theta}$ for all $n$.