What is a randomized controlled trial (RCT)?

The CONSORT 2025 explanation says that in random allocation each participant’s chance of receiving each intervention is known before assignment, yet the actual allocation is decided by chance and cannot be predicted. The US National Library of Medicine’s definition of the RCT publication type lists at least one test and one control treatment, concurrent enrollment and follow-up of both groups, and treatments chosen by a random process. CONSORT mentions the comparator as either an active treatment or a placebo.

According to CONSORT, randomization rests on two steps: generating an unpredictable allocation sequence and keeping it hidden from the investigators who enroll participants (allocation concealment). When properly implemented, this removes selection bias at entry into the trial. Blinding, or masking, is separate: it withholds the assigned intervention from people who could be influenced by knowing it, such as participants, care providers and outcome assessors, and CONSORT adds that it is not always appropriate or feasible.

CONSORT 2025 states that proper random assignment does not guarantee similar groups at baseline, and that baseline differences are due to chance rather than bias. Trials are reported with the CONSORT 2025 checklist of 30 items plus a participant flow diagram, which supersedes the 2010 version. ICMJE recommends that journals require a trial to be entered in a public registry no later than the first participant’s consent.