Definition
Control group: the group of participants in a clinical trial or experiment who do not receive the experimental treatment. Instead, they receive a placebo, no treatment, the standard treatment or a different dose. Because both groups are otherwise handled the same way, comparing their outcomes shows how much of the observed effect the treatment itself causes.
- The control group is the benchmark: without it, an improvement could come from the placebo effect, the natural course of the disease or chance.
- ICH E10 describes five main types: placebo, no-treatment, dose-response, active and external (historical) control.
- Participants are usually randomly assigned to treatment and control groups and, where possible, blinded.
Updated October 15, 2026
Learning clinical trial design? Control groups, randomization and blinding are part of the VIARES Clinical Research Associate course.
Contents
- What is a control group?
- Importance of control groups
- Types of control groups
- Control group vs experimental group
- Randomization and blinding
- Example: a placebo-controlled add-on trial
- Ethics: when is a placebo control acceptable?
- Control groups in observational study designs
- Common pitfalls
- What a CRA checks in the control arm
- FAQ
- Sources
- Where you learn this at VIARES
A new drug lowers blood pressure in patients who take it. Does that prove it works? Not yet: pressure can also fall because patients expect to improve. Only a control group, followed under the same conditions, separates the drug’s true effect from everything else.
What is a control group?
To define control group in one sentence: it is the comparison group in a study, the participants who do not get the intervention being tested. The NCI Dictionary of Cancer Terms describes it as the group that does not receive the new treatment, compared with the group that does.
The word “control” comes from the scientific control: everything that could influence the outcome is kept the same, so the only systematic difference is the treatment. Protocols also say control arm, comparator arm or standard of care arm.
ICH E10 distinguishes a concurrent control, followed at the same time as the test treatment group, from an external control, such as patients treated elsewhere or earlier. Concurrent, randomized controls are the basis of the randomized controlled trial, whether a pharmaceutical or biotechnology sponsor or an academic clinical and translational research team runs it.
Importance of control groups in a clinical trial
Patients in a clinical trial change for many reasons unrelated to the test treatment. Using control groups exposes these influences, because they affect both groups equally.
The placebo effect and why a placebo arm helps
- The placebo effect: expectations alone can improve symptoms, especially subjective ones such as pain or mood. Placebo responses are notably high in many neurological and psychiatric conditions (see placebo effect).
- The natural course of the disease: many health conditions improve, fluctuate or worsen on their own.
- Regression to the mean: patients are often enrolled when a value is unusually high, and the next measurement tends to be closer to their average.
- Other influences: extra attention, closer follow-up and confounding variables.
The difference between the groups, tested with a pre-specified statistical analysis, estimates treatment efficacy. Regulators require it: under 21 CFR 314.126, an adequate and well-controlled study must use a design that permits a valid comparison with a control.
Types of control groups: choosing the type of control group
ICH E10 and 21 CFR 314.126 list the same main types. The right choice depends on the disease, on whether an effective standard treatment exists, and on what the trial must show.
| Type (ICH E10) | What the control group receives | Typical use |
|---|---|---|
| Placebo concurrent control | An inactive look-alike, such as a sugar pill or saline injection | Showing a new drug works, when withholding treatment is acceptable |
| No-treatment concurrent control | Nothing; usually open-label | When a placebo is impossible, e.g. some surgical interventions |
| Dose-response concurrent control | A different dose of the same drug (sometimes plus a placebo arm) | Finding the dose range in Phase II |
| Active (positive) concurrent control | An approved, effective standard treatment as comparator | Superiority or non-inferiority trials when placebo would be unethical |
| Add-on design (placebo on top of standard treatment) | Standard treatment plus placebo (treatment group: standard plus new drug) | Testing a new medication without withholding existing care |
| External (historical) control | No concurrent group; patients from earlier studies or records | Rare diseases with a predictable course and dramatic effect; weakest against bias |
Control group vs experimental group: treatment and control groups compared
What is the difference between control group and experimental group? The experimental group (treatment or test group) receives the new intervention; the control group receives the comparison. Everything else is kept the same for control and experimental groups.
| Experimental group | Control group | |
|---|---|---|
| Receives | The experimental treatment (test treatment) | Placebo, no treatment, standard treatment or another dose |
| Role | Shows what happens with the intervention | Shows what happens without it: the benchmark |
| Assignment | Same eligibility criteria; participants randomly assigned to either group | |
| Measured on | The same endpoints, at the same visits, with the same methods | |
Control vs treatment group, test and control groups, experimental and control groups: the wording varies between medicine, psychological research and marketing A/B tests, but the logic is the same. Our experimental group article covers the treatment side in detail.
How participants are assigned: randomization and blinding
The comparison is only fair if both groups are similar at the start. Randomization achieves this: each participant is randomly assigned, so neither investigator nor participant chooses. ICH E9 describes randomization as the basis for avoiding bias in treatment assignment.
Blinding keeps the groups comparable during the trial, because knowing the assignment can change behavior and assessment. ICH E9 describes the levels:
- Single-blind: the participant (or, less often, the investigator) does not know the assignment.
- Double-blind: neither participants nor the investigators and sponsor staff involved in treatment or evaluation know it. A double-blind study with a placebo control group is the classic design.
- Double-dummy: when two drugs look different, each participant takes one active drug and a placebo matching the other, so both arms look identical.
Example from trial practice: a placebo-controlled add-on trial
Hypothetical example with invented numbers. A sponsor tests a new medication for high blood pressure. All 300 participants keep their standard treatment and are randomized 1:1, double-blind: 150 add the new drug, 150 a matching placebo. The primary endpoint is the change in systolic blood pressure at 12 weeks.
Blood pressure fell by 14 mmHg in the treatment group and 6 mmHg in the control group. The 6 mmHg happened without the new drug, so the estimated treatment effect is the difference: 8 mmHg. A single-arm study would have reported 14 mmHg, overstating the benefit by 75 %.
Ethics: when is a placebo control acceptable?
Giving a sugar pill instead of treatment raises ethical concerns. The Declaration of Helsinki of the World Medical Association (WMA) requires a new intervention to be tested against the best proven intervention. Placebo or no intervention is acceptable only if no proven intervention exists, or if compelling, scientifically sound methodological reasons require it and participants face no additional risk of serious or irreversible harm.
ICH E10 takes a similar line: a placebo control is generally unethical when an available treatment is known to prevent serious harm, such as death or irreversible morbidity. Then an active control or an add-on design is used.
Through informed consent, every participant learns that they may receive placebo. An ethics committee or IRB evaluates whether the choice of control is justified before the trial starts.
Control groups in observational study designs
Can observational studies have control groups? Yes, but the researcher does not assign the treatment, so the meaning shifts. In an observational study:
- Cohort studies compare people exposed to a factor (for example, a medication or smoking) with an unexposed comparison group, and follow both for outcomes.
- Case-control studies start from the outcome: “cases” have the disease, “controls” are similar people without it. Here, controls are defined by the outcome, not by treatment.
Without randomization, observational comparisons are more exposed to confounding.
Common pitfalls with control groups
- No control group: a before-and-after result cannot separate the treatment from background change.
- Selection bias: if staff choose the groups, they differ before treatment starts.
- Unblinding: side effects or careless drug handling reveal the assignment and bias assessments.
- Contamination: control participants obtain the experimental treatment elsewhere, shrinking the difference.
- Historical controls: diagnosis, standard care and populations change over time.
FAQ
What is a control group in an experiment?
A control group in an experiment is the group that does not receive the treatment or intervention being tested. It is kept under the same conditions as the treatment group, so researchers can compare outcomes and see whether the treatment caused a difference. In clinical trials, the control group usually receives a placebo, the standard treatment or no treatment.
How do you identify the control group?
Ask which group does not receive the intervention under investigation. That group is the control group; the group that receives the new treatment is the experimental group. In a clinical trial protocol, the control arm is named explicitly, for example as the placebo arm, the active comparator arm or the standard of care arm.
What does a lack of control group mean?
A lack of control group means there is nothing to compare the results with, so you cannot tell whether a change was caused by the treatment or by the placebo effect, the natural course of the disease, regression to the mean or chance. Single-arm studies without a control group can generate hypotheses, but they rarely prove that a treatment works.
Should treatment and control groups be the same size?
Not necessarily, but equal groups are the most common choice. A 1:1 allocation usually gives the highest statistical power for a given total number of participants. Trials sometimes use 2:1 or 3:1 so that more participants receive the new treatment or more safety data are collected, which requires a somewhat larger total sample size.
Is a control group positive or negative?
It can be either. A negative control receives no active treatment, such as a placebo or no treatment, and is expected to show no treatment effect. A positive control receives a treatment already known to work, such as an approved standard drug. In clinical trials, a positive control is called an active control.
Sources
- ICH E10: Choice of Control Group and Related Issues in Clinical Trials (ICH guideline)
- European Medicines Agency: ICH E10 Choice of control group in clinical trials (scientific guideline)
- ICH E9: Statistical Principles for Clinical Trials (ICH guideline)
- ICH E6(R3): Guideline for Good Clinical Practice (ICH guideline)
- 21 CFR 314.126: Adequate and well-controlled studies (eCFR)
- World Medical Association: Declaration of Helsinki, Ethical Principles for Medical Research Involving Human Participants
- National Cancer Institute: Definition of control group (NCI Dictionary of Cancer Terms)
Where you learn this at VIARES
Section 6, “Clinical Trial Design & Clinical Research Roles and Responsibilities”, of the VIARES Clinical Research Associate course covers the study protocol and trial design; its assignment asks you to explain stratified block randomization, double-dummy, cross-over and factorial designs and how they reduce bias. Section 10, “Investigational Medicinal Product Management”, covers comparators and placebo as investigational products, randomization and unblinding documentation, and drug accountability. VIARES also offers a free Micro Certificate CONTROL GROUP with a knowledge check (unlimited attempts, 80 % pass mark).
