Flight flutter testing is mandatory for aircraft certification; however it is a risky, lengthy and costly process with ever-growing demand for shorter and cheaper as well as online testing tools. This research has dealt with two key issues and the methods developed have been validated on simulated aeroelastic data sets. Firstly, the flutter boundaries are evaluated using statistical confidence criteria based on Least Squares statistics and eigenvalue perturbation theory rooting from noise statistics in the response data from a single flight test outdoing the need for Monte Carlo simulations or repeated tests. The approach is employed on three different flutter prediction tools: Damping Extrapolation method, Flutter Margin method and Envelope Function method. Results from the first two methods showed widening of the confidence bounds with increasing noise whilst the third method remained robust to it. The other research problem involves the development of an online flutter prediction tool using the Envelope Function method. The original approach is extended by making direct use of the chirp response to predict the flutter speed and is applicable to both offline and online versions.