This work concentrates on the mathematical modelling of marine vehiclemanoeuvring through the use of system identification. unlike previous research in this field which concentrates mainly on the measurement of hydrodynamic coefficients involving complex nonlinear models, this project intendsto develop alternative linear models that are simpler but will still be useful for the design of adaptive controllers for navigational autopilots.Modelling is performed through the use of computer programs on the manoeuvring of a tanker, freighter and minesweeper based on the determininstic auto-regressive moving average (DARMA) model structure. The input/output data ofrudder angle versus yaw angle used in the identification is simulated bysolving empirical state-space equations. The parameters in the manoeuvring models are then estimated by using the Recursive Least Squares algorithmbefore the models are finally tested by model validation techniques.Results of the correlation tests on model validity indicate that linear models are inadequate to fully describe manoeuvring in the actual sea conditionsbut are sufficient when used in a closed-loop navigational control system.