An arm of a clinical trial is a group of patients receiving a specific treatment (or no treatment). Trials involving several arms, or randomized trials, treat randomly-selected groups of patients with different therapies in order to compare their medical outcomes. Experimental arms, which receive an experimental drug, are compared with control arms, which can receive an active comparator (another therapy used to treat the same condition as the experimental therapy), a placebo comparator (an inactive therapy), a sham comparator (an inactive therapy made to look identical to the active therapy), or no intervention. Some clinical trial designs, such as the lung cancer master protocol design, allow for several experimental drugs to be tested simultaneously.

Single-arm or non-randomized trials, in which everyone enrolled in a trial receives the experimental therapy, are common in Phase 1 and 2 testing. While Phase 3 trials are frequently randomized in order to provide more precise data on safety and effectiveness outcomes, a Phase 3 trial might use a single arm if a small target population makes conducting a randomized trial impractical. In these cases, researchers must use endpoints like response rate that can demonstrate clinical impact in the absence of direct comparison data.