Growth rates are widely employed in the field of agriculture as these have important policy implications. The usual parametric approach for growth rate analysis is to assume multiplicative error in the underlying nonlinear geometric model and then fit the linearized model by ‘method of least squares’. The deficiencies of this approach have been highlighted. It has been argued that nonlinear estimation procedures should be employed for fitting the model and then only the growth rates should be computed. A methodology has been discussed to compute the compound growth rate by using growth models, viz. monomolecular, logistic and Gompertz. Further, as an illustration, the total foodgrain production of our country during the period 1980 to 2001 has been considered and its growth rate has been computed.