In this thesis, localization of the underwater sound sources using matched field processing technique is considered. Localization of the underwater sound sources is one of the most important problems encountered in underwater acoustics and signal processing. Many techniques were developed to localize sources in range, depth and bearing angle. However, most of these techniques do not consider or only slightly takes into account the environmental factors that dramatically effect the propagation of underwater sound. Matched field processing has been developed as a technique that fully considers the environmental factors. Matched field processing has proven to be successful in many applications such as localization of sources in range and depth, the determination of environmental parameters, and the evaluation of model accuracies. In this study, first a comparative analysis of narrowband matched field processors is given. Namely four main processors: Bartlett processor, Minimum Variance Distortionless Response (MVDR) processor, MVDR with neighboring location constraints and MVDR with environmental perturbation constraints are compared in terms of their probability of correct localization under certain environmental conditions. Secondly, a performance assesment for the most common broadband matched field processors is made. The correct localization performances for incoherent broadband matched field processor, Tolstoy/Michalopoulo&#039 / s coherent matched field processor and broadband matched field processor with environmental perturbation constraints is given for certain environmental conditions. Finally, a new weighting approach to combine data for broadband matched field processing is introduced. The fact that information from different frequencies may have different reliability depending on the environmental conditions is considered to develop a weighting scheme. It is shown that a performance gain compared to existing processors can be achieved by using the weighting scheme introduced in this study.