This paper examines the efficiency of various methods of calibrating a rainfall-runoff model. The model used is a 12 parameter version of the Boughton model which has been developed for large tropical basins. Attempts were made to improve the efficiency of calibration in three areas: selection of the best nonlinear programming algorithms; reduction of the number of objective functions required for calibration; and simplification of the model structure. The best algorithms were found to be those of Powell, Rosenbrock, and the simplex method of Nelder and Mead. The Davidon method did not perform well. The number of objective function evaluations can be reduced by performing a sensitivity analysis on the model and selecting a small group of parameters which are not interdependent and which the objective function is sensitive to. This may yield a substantial reduction in the computer time required to calibrate the model. Simplification of the model structure can also yield substantial savings, especially where it removes calculations which are redundant and reduces the number of model parameters.