If you have any random effects or repeated measures in your multilevel model then you have to decide upon the covariance structure of your data. If you have random effects and repeated measures then you can specify different covariance structures for each. The covariance structure simply specifies the form of the variance–covariance matrix (a matrix in which the diagonal elements are variances and the off-diagonal elements are covariances). There are various forms that this matrix could take and we have to tell SAS what form we think it does take.

SAS has 23 different covariance structures that we can use,Which Covariance Structure Should We Choose?