In this paper a method based on wavelet transform (WT) and particle filtering (PF) for estimation of single trial event-related potentials (ERPs) is presented. The method is based on recursive Bayesian mean square estimation of wavelet coefficients of the ERPs, using PF as the estimator. Simulation results are provided to demonstrate the superior performance of PF over Kalman filtering (KF) for non-Gaussian and non-linear electroencephalography (EEG) signals. The methods were also applied to the real data in an odd-ball paradigm to explore the changes in the P300 component from trial to trial.