Kaplan-Meier Plotter
Paste survival data with censoring and get product-limit survival curves, median survival, censoring marks, and a log-rank test between two groups. Download a publication-ready SVG.
Drag & drop a file or
CSV, TSV, Excel (.xlsx/.xls) - max 5000 rows
| Number at risk | 0 | 5 | 10 | 15 | 20 |
|---|---|---|---|---|---|
| Treatment | 9 | 9 | 5 | 3 | 2 |
| Control | 9 | 3 | 1 | 0 | 0 |
Estimating Survival When Not Everyone Has the Event
Survival analysis exists to handle a problem ordinary averages cannot: at the end of a study, many subjects have not yet had the event, so their true event time is unknown. The Kaplan-Meier estimator, introduced by Kaplan and Meier (1958), solves this with the product-limit method, multiplying together the conditional probabilities of surviving each successive event time. Subjects with unobserved event times are censored, counted as at risk until they leave, so their partial information is used rather than discarded.
The resulting step curve reads directly: its height at any time is the estimated probability of surviving beyond that time, it drops only at event times, and small ticks mark censored subjects. The most common summary drawn from it is the median survival, the time the curve first crosses 0.5. When most subjects remain event-free at the end of follow-up the curve never reaches 0.5, and the honest report is that median survival was not reached.
Comparing two groups needs a formal test, because two curves can cross or separate by chance. The log-rank test accumulates, at every event time, the difference between observed and expected events under the null hypothesis of identical survival, producing a chi-square statistic and a p-value. This tool computes it automatically for two-group data, so a treatment-versus-control comparison returns both the curves and the significance in one step.
Kaplan-Meier and the log-rank test are the standard unadjusted analysis; they describe survival and compare groups but cannot adjust for other variables. That is the role of Cox proportional-hazards regression, the multivariable follow-up. If your curves come from a published figure rather than raw data, digitize them first with the survival curve digitizer, and for a full time-to-event analysis with adjusted hazard ratios, the biostatistics consulting service runs the Cox models and reporting.
A Worked Example: The Product-Limit Estimate Step by Step
Follow five subjects with event times 2, 3, 8 and 9, and one subject censored at time 5. At each event the survival probability is multiplied by one minus the fraction of at-risk subjects who had the event; a censored subject does not lower the curve but does leave the at-risk set. The table below is exactly what this tool computes.
| Time | At risk | Events | Step factor | Survival S(t) |
|---|---|---|---|---|
| 2 | 5 | 1 | 1 − 1/5 | 0.80 |
| 3 | 4 | 1 | 1 − 1/4 | 0.60 |
| 5 | 3 | 0 (censored) | no change | 0.60 |
| 8 | 2 | 1 | 1 − 1/2 | 0.30 |
| 9 | 1 | 1 | 1 − 1/1 | 0.00 |
The median survival is 8, the first time the curve reaches or drops below 0.5. The censoring at time 5 is the instructive step: the curve stays flat at 0.60, but that subject is removed from the at-risk count, so the next death at time 8 divides by 2 rather than 3, giving 0.60 times (1 − 1/2) = 0.30. That is how censored observations still shape the curve without being counted as events. The same product-limit logic, applied to the two arms of the built-in leukemia-remission example, yields median remission of 13 versus 3 and a log-rank test of chi-square 9.71 at one degree of freedom, p = 0.0018.
Common Mistakes in Kaplan-Meier Analysis
- Mishandling censored subjects. Dropping censored subjects, or counting them as events, biases the survival estimate. They must stay in the at-risk set until their censoring time, which is exactly how the product-limit estimator treats them.
- Reversing the event code. This tool reads 1 as the event and 0 as censored. Files that store 1 for censored invert the meaning, so every curve comes out wrong. Confirm the coding before importing, and note that a dedicated censoring column is not read here for this reason.
- Judging a difference by eye. Two curves that look separated may not differ significantly, and overlapping curves can still differ. Use the log-rank test, not the visual gap, to decide.
- Over-reading the tail. Late in follow-up only a few subjects remain at risk, so the curve becomes unstable and a single event causes a large step. Treat the right-hand end of the curve with caution and note where the at-risk count is small.
- Using Kaplan-Meier when you need adjustment. The estimator and log-rank test are unadjusted. To account for covariates or estimate a hazard ratio, follow up with Cox proportional-hazards regression rather than reading everything from the curves alone.
How to Use This Plotter
Paste your data
One row per subject with the follow-up time, event indicator (1 event, 0 censored), and an optional group.
Read the curves
A product-limit curve per group, with ticks for censored subjects and the median survival reported.
Compare groups
With two groups, the log-rank test reports whether the survival curves differ significantly.
Export
Download the survival plot as a publication-ready SVG.
Next step
Want a PhD to run the full survival analysis?
Cox proportional-hazards models, hazard ratios, and publication-ready figures with a reproducible methods section, handled by a PhD statistician.
Our promise: Free re-run and re-write if reviewers question the analysis or reporting.
Timeline
Most projects deliver in under 2 weeks. We confirm an exact date in your quote.
If reviewers push back
If reviewers question the analysis, assumptions, or reporting, we re-run and re-write free.
Confidentiality
NDA available on request before any project discussion. Your data, study design, and manuscript stay private either way.
Want a PhD methodologist to handle the whole project?
Get a complete survival analysis with Cox models, hazard ratios, and publication-ready figures by a PhD statistician. Free re-run and re-write if reviewers question the analysis or reporting. Pay only after you approve your quote.
Frequently Asked Questions
What is a Kaplan-Meier curve?
A Kaplan-Meier curve is a step function that estimates the probability of surviving past each point in time from data where some subjects have not yet had the event. It uses the product-limit estimator: at each time an event occurs, the survival probability is multiplied by one minus the fraction of at-risk subjects who had the event. The curve starts at one and steps downward at each event time, giving a picture of survival over the whole follow-up rather than a single summary number.
What is the Kaplan-Meier curve used for?
It is used to estimate and display the probability of surviving over time when the outcome is time until an event, such as death, relapse, or device failure, and some subjects are censored. Beyond describing survival in a single group, it is used to compare two or more groups, usually with a log-rank test, and to read off summaries like the median survival time. It is a staple of clinical trials, oncology outcomes, and reliability engineering wherever time-to-event data with incomplete follow-up arise.
How do you read a Kaplan-Meier curve?
The height of the curve at any time is the estimated probability of surviving beyond that time. A curve that stays high indicates good survival, and a steep drop marks a period of high event risk. When two curves are plotted, the one that stays higher has better survival, and the gap between them shows the size of the difference. Small vertical ticks mark censored subjects, those who left the study or had not had the event at last contact.
When should I use Kaplan-Meier analysis?
Use it when your outcome is the time until an event and some subjects are censored, and you want an unadjusted estimate of survival or a comparison between a small number of groups. It is the right first analysis for a two-arm trial or an observational comparison by exposure. Reach for something else when you need to adjust for covariates or model continuous predictors, where Cox proportional-hazards regression is appropriate, or when there is no censoring at all, where simpler summaries may suffice.
What is the difference between Kaplan-Meier and the survival curve?
A survival curve is the general concept: the function that gives the probability of being event-free over time. Kaplan-Meier is one specific way to estimate that curve from observed data, the non-parametric product-limit estimator, which makes no assumption about the shape of the underlying survival function. So a Kaplan-Meier curve is a survival curve, but not the only kind; parametric models such as Weibull or exponential, and model-based curves from Cox regression, are alternative estimates of the same underlying survival function.
What is censoring in survival analysis?
Censoring occurs when a subject's event time is not fully observed, most often because the study ended or the subject was lost to follow-up before the event happened. These subjects still contribute information, because they were known to be event-free up to their censoring time, and the Kaplan-Meier estimator uses that by keeping them in the at-risk count until they are censored. Ignoring censoring, or treating censored subjects as events, would bias the survival estimate.
What is the log-rank test?
The log-rank test compares the survival experience of two or more groups by accumulating, at each event time, the difference between the observed number of events in a group and the number expected if the groups had identical survival. The summed differences form a chi-square statistic; a small p-value indicates the survival curves differ more than chance would allow. This tool runs the log-rank test automatically when the data contain exactly two groups.
What is median survival?
Median survival is the time at which the survival curve first drops to 0.5, meaning half the subjects have had the event and half have not. It is the most common single-number summary of a survival curve because it does not require the curve to reach zero. When more than half the subjects are still event-free at the end of follow-up, the curve never reaches 0.5 and the median survival is reported as not reached.
What is the difference between Kaplan-Meier and Cox regression?
Kaplan-Meier is a non-parametric description of survival in one or a few groups and answers what the survival looks like, but it cannot adjust for other variables. Cox proportional-hazards regression models the hazard as a function of covariates, giving adjusted hazard ratios and handling continuous predictors. Kaplan-Meier curves and the log-rank test are the standard unadjusted analysis; Cox regression is the multivariable follow-up.
Related Analysis Tools
To recover survival data from a published figure, the survival curve digitizer extracts the coordinates. For sample-size planning of a survival study, the sample size calculator handles time-to-event outcomes, and to convert a test statistic to a p-value the p-value calculator covers the chi-square distribution. For a full survival analysis with adjusted hazard ratios, the biostatistics consulting service runs the Cox models.
Reviewed by
Dr. Sarah Mitchell
Dr. Sarah Mitchell holds a PhD in Biostatistics from Johns Hopkins Bloomberg School of Public Health and has over 15 years of experience in systematic review methodology and meta-analysis. She has authored or co-authored 40+ peer-reviewed publications in journals including the Journal of Clinical Epidemiology, BMC Medical Research Methodology, and Research Synthesis Methods. A former Cochrane Review Group statistician and current editorial board member of Systematic Reviews, Dr. Mitchell has supervised 200+ evidence synthesis projects across clinical medicine, public health, and social sciences. She reviews all Research Gold tools to ensure statistical accuracy and compliance with Cochrane Handbook and PRISMA 2020 standards.
The Tool Is Free. The Full Analysis of Your Biological Data? We Handle That.
Our PhD statisticians run the complete pipeline: differential expression with multiple-testing correction, survival modelling, dimensionality reduction, and publication-ready figures with a reproducible methods section. Constant pricing, most projects delivered in under two weeks.
Our promise: Free pipeline re-run and figure revisions if reviewers push back.