The presented paper deals with a constrained optimisation technique for the calibration of elasto-plastic model parameters in a rational and objective manner. The procedure consists in finding a set of model parameters which minimise the difference between the experimental data and the numerical simulations defined by an objective function. For this purpose, an optimisation routine, termed ParaID, has been developed which combines the quasi-Newton and stochastic methods. The optimisation technique was employed to calibrate a multi-mechanism elasto-plastic constitutive model. Using the results of three isotropically consolidated drained triaxial compression tests, a comparison between numerical and experimental results clearly shows the capability of the optimisation procedure to determine the model parameters correctly.