How to search ClinicalTrials.gov, a practical guide for professionals

ClinicalTrials.gov is run by the US National Library of Medicine. It holds registration records for interventional and observational studies conducted in the United States and in many other countries, submitted by sponsors and investigators.

Each record describes a study at registration and is updated over time: its design, arms and interventions, eligibility criteria, outcome measures, locations, status and key dates. For many trials the record also contains posted summary results.

• Registration of many interventional trials is required by US law, and most medical journals require prospective registration in a public registry as a condition of publication.
• Records are written by the sponsor or investigator. The registry runs quality control checks, but it does not verify scientific accuracy.
• Not every trial in the world is here. Trials registered only in other national or regional registries need a separate search, which we cover in our guides to the EU clinical trial register and the international clinical trials registry platform.

The search form has separate fields for different parts of your question. Putting a term in the right field is the single biggest improvement you can make.

Condition or disease

Enter the disease, disorder or condition studied, for example a tumour type or a cardiovascular condition. The registry maps many condition terms to related synonyms, but coverage is not complete, so check alternative names.

Intervention or treatment

Enter the drug, device, procedure or behavioural intervention. Include the generic name, brand names and development code names, because different records use different names for the same product.

Other terms

This field searches across the whole record, including titles, descriptions and identifiers. Use it for concepts that do not fit the other fields, such as a mechanism of action, a biomarker, an outcome or a sponsor study code. It is also where you can paste an NCT number.

Location

Search by country, state or city, and on the newer interface by distance from a location. Location filters show where a study has sites listed, not where most participants were enrolled.

If you structure your question first with the PICOS framework, the population maps to condition, the intervention to intervention, and outcomes and design to other terms and filters.

Filters work on structured fields, so they are more precise than adding words to a text field.

• Study status: not yet recruiting, recruiting, enrolling by invitation, active not recruiting, completed, suspended, terminated, withdrawn and unknown status. For feasibility you usually want recruiting and not yet recruiting. For evidence on outcomes you usually want completed, and often terminated as well.
• Phase: early phase 1, phase 1, phase 2, phase 3 and phase 4. Some trials are registered as combined phases, such as phase 1 and phase 2, so check how combined phases are handled in your filter.
• Study type: interventional, observational and expanded access. Observational studies include cohorts, case control studies and registries.
• Results: studies with results posted or without results. This is the quickest way to find trials whose summary outcomes are available in the registry.
• Eligibility: age groups, sex and whether healthy volunteers are accepted.
• Funder type: industry, NIH, other US federal agencies, or other funders such as universities and hospitals.
• Dates: study start, primary completion, completion, first posted, results first posted and last update posted.

Every study in ClinicalTrials.gov has a unique identifier in the form NCT followed by eight digits. It is assigned at registration and does not change.

• Searching by NCT number is the most reliable way to find a specific trial, so record the number whenever you cite a trial.
• Many publications list the NCT number in the abstract or methods section, which lets you link a paper to its registry record and compare reported outcomes with registered ones.
• Secondary identifiers in the record, such as sponsor protocol numbers or EudraCT and EU CT numbers, help you find the same trial in other registries.
• Every record has a history of changes, so you can see what the registered primary outcome was at the start and whether it changed later.

When summary results are posted, the record gains a results section in a standard structure. Reading it well takes practice.

• Participant flow shows how many participants started, completed and withdrew in each arm, with reasons.
• Baseline characteristics describe the enrolled population by arm, including age, sex and other measures the sponsor reports.
• Outcome measures list each primary and secondary outcome with its time frame, the results per arm and, where provided, the statistical analysis.
• Adverse events list serious and other adverse events by arm, with the time frame and how events were collected.
• Limitations and caveats, and sometimes a link to the publication, appear at the end.

Posted results are summary data provided by the sponsor. They are not peer reviewed in the way a journal article is, and they may differ from what was later published. Comparing the two is a standard step in evidence review.

Most missed trials and misread records come from a handful of predictable problems.

• Synonyms and naming. A drug can appear under a code name in early records and a generic or brand name later. Conditions are described in many ways. Search each name, or use a tool that searches by meaning.
• Status lag. Status is updated by the sponsor, so a trial shown as recruiting may have stopped, and a completed trial may still show active not recruiting. Check the last update posted date and the last verified date.
• Unknown status. Records whose status has not been verified for a long time after their expected completion are marked as unknown status. Treat them with caution in a feasibility count.
• Duplicates across registries. A multinational trial may be registered in ClinicalTrials.gov, the EU register and a national registry under different identifiers. Counting each record separately inflates the landscape.
• Results not posted does not mean no results. Many trials publish in journals without posting registry results, and some post results years after completion.
• Registered outcomes versus reported outcomes. When the primary outcome changed during the trial, the change is visible in the record history.
• Too many words in one field. Long phrases in the condition field can return fewer results than expected. Start broad, then add filters.

The registry interface is designed for finding studies, not for answering questions about them. An AI layer that is grounded in registry data can shorten the step between search and answer.

• Semantic search finds trials by meaning, so mechanisms, drug classes and synonyms are covered without a long Boolean string.
• One question searches ClinicalTrials.gov together with PubMed, PubMed Central, the EU Clinical Trials Register and the WHO ICTRP, which helps you spot trials registered in several places.
• Structured fields such as phase, population, interventions, endpoints, sponsor, status, enrolment and results posted land in one sortable clinical trial table.
• A cited answer summarises what the matching trials report, with each sentence linked to its NCT number or PMID.

ClinicalEvidence does all of this from a daily refreshed index, verifies every citation before you see it and says plainly when no evidence matches. Read more on our page about ClinicalTrials.gov search, or set up clinical trial alerts to follow new registrations and posted results for a saved question.