The statistical analysis of the multiply imputed data involved repeated analysis followed by parameter pooling. Rubin’s rules apply to a wide variety of quantities, especially if these quantities are transformed toward normality. Dedicated statistical tests and model selection technique are now available. Although many techniques for complete data now have their analogues for incomplete data, the present state-of-the-art does not cover all. As multiple imputation becomes more familiar and more routine, we will see new post-imputation methodology that will be progressively more refined.