Analysis of Covariance is a standard technique for the analysis of randomized clinical trials with continuous outcome measured at baseline and follow-up. The treatment assignment is entered as a categorical predictor and the baseline value of the outcome measure as a continuous covariate. The response variable may be taken as either the value of the outcome measure at follow-up or its change from baseline to follow-up. For measurements that are subject to high degree of variability it may be advisable to repeat baseline and/or follow-up assessments. Repetition of the follow-up assessment is generally more advantageous than repetition of the baseline assessment, but the latter can still be valuable. In the context of linear models, calculation of the resulting gain in power (or reduction in total sample size needed to achieve a given power) is straightforward. For nonlinear models, for example when the outcome is categorical, exact results are generally not available but approximate results may be obtained by simulation. We report investigations of these issues in the context of a clinical trial in Huntington's disease.