TimelyScholarFree Tools › ROBINS-I V2 Tool
Record ROBINS-I Version 2 judgments across its six bias domains for each result from a non-randomized study, see the overall risk of bias suggested by the V2 algorithm, and download traffic light and summary plots.
FreeNo account neededYour data stays in your browser
Does your protocol specify the original 2016 version, with seven domains? Use the ROBINS-I (2016) tool →
Answer the signaling questions in the official tool, then record your judgments here: ROBINS-I V2 tool and template (riskofbias.info).
| Study, result | D1 | D2 | D3 | D4 | D5 | D6 | Overall | |
|---|---|---|---|---|---|---|---|---|
Your entries are saved in this browser only. Nothing is uploaded.
ROBINS-I V2 sets the overall risk of bias to the worst domain judgment. You may raise it: several domains at moderate can justify serious, and several at serious can justify critical; give your reason. If the triage questions (part B) show the result is at critical risk of bias, choose Critical for the overall judgment and leave the domains blank.
For each result, note the outcome, the target trial the study emulates, and whether it estimates the intention-to-treat or per-protocol effect (V2 parts A and C).
If part B shows the result is at critical risk of bias, add a row, choose Critical for the overall judgment and leave the domains blank.
Answer the signaling questions in the official V2 document, whose algorithms propose each domain's judgment, then record them here: one row per result.
The worst domain is suggested; raise it if several domains share problems. Download the plots as PNG or SVG and the table as CSV.
ROBINS-I assesses the risk of bias in a specific result from a non-randomized study that compares an intervention strategy with a comparator strategy, such as a cohort study. Each result is judged against a target trial: the hypothetical randomized trial the study is trying to emulate.
Version 2 (ROBINS-I V2), most recently updated on riskofbias.info on 20 November 2025, has six domains: confounding, classification of interventions, selection of participants into the study (or analysis), missing data, measurement of the outcome, and selection of the reported result. It dropped the 2016 domain on deviations from intended interventions, and the confounding domain now has variants for the intention-to-treat effect and the per-protocol effect.
Algorithms map the answers to the signaling questions onto proposed judgments of low, moderate, serious or critical risk of bias. 'No information' is now an answer to signaling questions rather than a judgment. For confounding, the best judgment is low risk of bias except for concerns about uncontrolled confounding, because uncontrolled confounding cannot be ruled out in an observational study.
V2 is still labeled a draft and may change. If your protocol specified the 2016 version of ROBINS-I, keep using that version, with our ROBINS-I (2016) tool, and report which one you used.
V2 adds algorithms that turn answers to signaling questions into proposed judgments, 'strong' and 'weak' yes and no answers, a triage section that can rate a result as critical straight away, and two variants of the confounding domain. It drops the domain on deviations from intended interventions and the 'no information' judgment, and renumbers the other domains.
D1 confounding, D2 classification of interventions, D3 selection of participants into the study (or into the analysis), D4 missing data, D5 measurement of the outcome, and D6 selection of the reported result.
V2 is the current version on riskofbias.info but is still labeled a draft. Use the version your protocol specifies, report which version you used, and apply the same version to every study in the review. Our ROBINS-I (2016) tool records the seven original domains.
There is no journal article for V2 yet. Cite the tool document: Sterne J, Higgins J, and the ROBINS-I V2 development group, ROBINS-I V2 assessment tool (for follow-up studies), 20 November 2025, available from riskofbias.info, alongside Sterne et al. (2016) for the original tool.
The official site provides an editable template for answering the signaling questions. This tool exports your judgments as a CSV file that opens in Excel, and can import a CSV with one row per result.
ROBINS-I is for studies of interventions (things people choose or are given). ROBINS-E is for studies of exposures, such as environmental or occupational exposures.