The demand for enhanced performance of production systems in terms of quality, cost and reliability is ever increasing while, at the same time, there is a demand for shorter design cycles, longer operating life, minimisation of inspection and maintenance needs. Experimental testing and system identification in operational conditions still represent an important technique for monitoring, control and optimization. The term identification refers in the present paper to the extraction of information from experimental data and is used to estimate operational dynamic parameters for machining systems. Such an approach opens up the possibility of monitoring the dynamics of machining systems during operational conditions, and can also be used for control and/or predictive purposes The machining system is considered nonlinear and excited by random loads. Parametric and nonparametric techniques are developed for the identification of the nonlinear machining system and their application is demonstrated both by numerical simulations and in actual machining operations. Discrimination between forced and self-excited vibrations is also presented. The ability of the developed methods to estimate operational dynamic parameters ODPs is presented in practical machining operations.