Abstract Previously it has been established that when mean relative growth rate (RGR) is estimated from growth data by the classical method of growth analysis, the computed variance of RGR is probably greater than it should be. By similar procedures, it is now demonstrated that the same is true of unit leaf rate (ULR). The experimental results indicate that the coefficients of variation of both RGR and ULR are similar in magnitude. In an ordinary growth experiment, analysed by the classical methods of growth analysis, assuming the equality of the coefficients of variation of RGR and ULR enables a simple calculation of the variance of ULR to be made. Thus, in such experiments, lower variances of both RGRs and ULRs are obtained than are given by the classical methods. These lower variances are probably closer to biological reality, and also increase the sensitivity of such experiments.