This work investigates the importance of experimental design on the accuracy and precision of estimates of the agglomeration, growth and nucleation kinetics for the gibbsite system. The work is motivated by the need to explain the differences in crystallisation kinetics parameters reported in the literature for this system. Experiments were conducted to estimate the kinetics under the same process conditions (e.g. temperature, hydrodynamics and composition) from three different crystalliser configurations, i.e. (1) batch; (2) semi-batch at constant composition with representative solids removal and (3) semi-batch at constant composition with classified solids removal. The differential method of Bramley, Hounslow, and Ryall (J. Colloid Interface Sci. 183 (1996) 155) modified For semi-batch crystallisation, was used to estimate the kinetics from each set of experimental data. Some differences in the kinetics estimates between the different configurations were observed. Dynamic simulation showed that each set of estimated kinetics was adequate for describing the experimental CSD's. The uncertainties in the agglomeration and growth rate estimates, obtained by Monte Carlo simulations, were large enough to explain the observed differences. The differences in the nucleation source term were attributed to differences in suspension density between the experiments. A sensitivity analysis found that operating conditions have a greater impact on the uncertainties in the estimated kinetics than the crystalliser configuration.