It is widely believed that baseline imbalances in randomized clinical trials must necessarily be random. Yet even among masked randomized trials conducted with allocation concealment, there are mechanisms by which patients with specific covariates may be selected for inclusion into a particular treatment group. This selection bias would force imbalance in those covariates, measured or unmeasured, that are used for the patient selection. Unfortunately, few trials provide adequate information to determine even if there was allocation concealment, how the randomization was conducted, and how successful the masking may have been, let alone if selection bias was adequately controlle d. In this article we reinforce the message that allocation details should be presented in full. We also facilitate such reporting by identifying and clarifying the role of specific reportable design features. Because the designs that eliminate all selection bias are rarely feasible in practice, our development has important implications for not only the implementation, but also the reporting and interpretation, of randomized clinical trials.