What Is a Randomised Controlled Trial (RCT) in Research?

What is a randomised controlled trial?

A randomised controlled trial (RCT) is a study in which researchers randomly assign participants to two or more groups, give each group a different intervention (or none), and compare the outcomes. One group might receive a new treatment, for example, while the control group receives a placebo, standard care, or nothing. Because assignment is random, the groups should be similar in every way except the intervention, which means any difference in outcomes can be attributed to the intervention rather than to pre-existing differences between people.

RCTs are described as "the gold standard for evaluating interventions because of their ability to minimise or avoid bias"^1,2. That phrase appears repeatedly in the methodological literature, and it matters: a well-run RCT is the strongest design we have for working out whether something actually causes an effect.

Why does randomisation matter?

The defining feature of an RCT is random assignment. When participants are allocated to groups by chance, factors that could influence the outcome, such as age, baseline health, motivation, or lifestyle, should be distributed roughly evenly across groups. That is what allows researchers to isolate the effect of the intervention itself.

Observational studies, by contrast, simply watch what happens to people who are already doing different things. If people who take a supplement also report better sleep, you cannot tell whether the supplement helped or whether supplement-takers differ in other ways, such as being more health-conscious overall. Randomisation breaks that link between the intervention and other characteristics, which is why RCTs can support causal claims while observational studies can usually only support associations.

How an RCT works in practice

A typical parallel-group RCT follows a straightforward sequence. Researchers define who is eligible, recruit participants, and randomly assign each person to a group. One group receives the intervention; the other receives a comparison. Both groups are then followed over the same period, and outcomes are measured in the same way. Finally, the results are analysed by comparing outcomes between groups.

The CONSORT (Consolidated Standards of Reporting Trials) statement, first published in 1996 and updated since, provides a checklist and flow diagram that authors use to report this process transparently³. The checklist covers the title, abstract, introduction, methods, results, and discussion, and includes 22 items selected because evidence indicates that failing to report them is associated with biased estimates of treatment effect³. The flow diagram shows how many participants were enrolled, allocated, followed up, and analysed in each group, which lets readers judge whether the analysis was done properly³.

What makes an RCT strong evidence?

Three features give a well-conducted RCT its strength. First, randomisation minimises selection bias, the systematic differences between groups that can distort results. Second, a control group provides a benchmark, so researchers can see what would have happened without the intervention. Third, when the trial is properly reported, readers can check exactly what was done and judge whether the findings are trustworthy.

That third point is less abstract than it sounds. Methodological analyses have found that inadequate reporting and design are associated with biased estimates of treatment effects, which is "seriously damaging to RCTs"^1,2. In other words, randomisation alone does not guarantee a reliable result. A poorly designed or poorly reported RCT can still mislead, which is why reporting standards such as CONSORT exist and why many leading medical journals endorse them^1,2.

Where do pilot and feasibility studies fit?

Before running a full RCT, researchers often run a smaller preparatory study. A feasibility study asks whether something can be done, whether researchers should proceed with it, and if so, how. A pilot study asks the same questions but also has a specific design feature: a future study, or part of a future study, is conducted on a smaller scale⁴. In other words, pilot studies are a subset of feasibility studies, rather than the two being mutually exclusive⁴.

This matters when you read about early research. A pilot study is not designed to prove that an intervention works. It is designed to test whether a larger trial is practical, for example whether enough people can be recruited, whether the intervention can be delivered as planned, and whether outcome measures work. If a news story cites a pilot study as proof of an effect, treat it with caution: the study was not built to answer that question.

RCTs are not the only credible design

RCTs are the gold standard, but they are also resource-intensive, time-consuming, and operationally complex to conduct⁵. That is one reason researchers are exploring alternatives. Target trial emulation is a framework for designing observational studies using routinely collected healthcare data so that they more closely resemble a target trial⁵. Artificial intelligence may eventually support parts of this process, from defining study populations to analysing outcomes, though researchers caution that AI should support rather than replace human judgement, and that bias, lack of transparency, and over-reliance on automated methods are real risks⁵.

The practical takeaway is not that RCTs are the only evidence worth reading. It is that they occupy a specific place: when you want to know whether an intervention causes an outcome, a well-conducted, well-reported RCT is the strongest single study design available. For more on how evidence fits together, see our plain-English guide to meta-analyses and our explainer on why correlation is not causation.

How to read an RCT critically

When you encounter a claim based on an RCT, a few checks will tell you how much weight to give it. Look for how participants were randomised and whether the groups were similar at the start. Check whether there was a genuine control group. Note how many people dropped out and whether the analysis included everyone who was randomised, the intention-to-treat principle that the CONSORT flow diagram is designed to reveal³. And check whether the trial was registered and reported against a standard checklist.

If the trial is small, short, or conducted in a population very different from you, its findings may not generalise. An RCT can be internally valid, meaning its result is trustworthy for the people studied, while still being of limited relevance to your situation. That distinction is often lost when a single trial becomes a headline. For more on that problem, see our piece on why so many wellness claims rest on a single small study.