Baseline characteristics should be measured before patients are randomized so that knowledge of the treatment assignment cannot influence reporting of baseline characteristics. It is important to monitor all patients with respect to the treatment, both the treatment that was assigned as well as the treatment eventually received (as all patients may not be treated according to the random assignment). Some patients may refuse the assigned treatment; in a pharmaceutical trial, some patients may not comply fully, taking only a portion of the assigned medication. Methods such as pill counts (or more high-tech approaches) are useful in assessing the level of compliance with the assigned treatment.

The "outcome" is the factor on which the assessment of treatment efficacy is based. In order to determine whether a new treatment is "better", the criterion on which it will be judged, the primary outcome, must be specified in advance. Similarly, other effects, both desirable and undesirable, are generally measured in order to gain a more complete picture of the benefits and risks of the new treatment.