In a repeated measures model each of n individuals receive each of k treatments. Usually a one-way ANOVA model is used under the assumption that the vectors of observations on the j-th individual \(Y_ j=(Y_{1j},...,Y_{kj})'\), say, are independently \(N\{\mu,\sigma^ 2A(\rho)\}\) distributed, where \(\mu =(\mu_ 1,...,\mu_ k)'\) is the mean of \(Y_ j\), \(\sigma^ 2\) is the common variance of \(Y_{ij}\) \((i=1,...,k)\) and A(\(\rho)\) is a correlation matrix with 1 in the diagonal and \(\rho\) in each of the off-diagonal terms. The purpose of this note is to point out that known results on inference from independent random samples \((\rho =0)\) about several normal means under order restrictions can simply be generalized to this repeated measures model (\(\rho\neq 0)\). The order restrictions on the parameter space \(R^ k\) of \(\mu\) are supposed to be of the form \(\mu_ i\leq \mu_ j\).