This thesis mainly discusses the theory and applications of an estimation technique called Bootstrap. The first part of the thesis focuses on the accuracy of Bootstrap in density estimation by comparing Bootstrap with another estimation technique called Normal approximation based on central limit theorem. The theoretical analysis on this issue shows that Bootstrap is always, at least as good as, and in some cases better than, the Normal approximation. This analysis has been supported by empirical analysis. Later parts of the thesis are devoted to the applications of Bootstrap. Two examples for these applications. Bootstrapping F-test in dynamic models and using Bootstrap in common factor restrictions have been extensively discussed. The performance of Bootstrap has been investigated separately and interpreted precisely. Bootstrap has worked well in F-test application, but it has been dominated by other tests such as Likelihood Ratio test, Wald test; in common factor restrictions.