The Walking Speed of Death: How Fast Does the Grim Reaper Actually Move?

Every year the BMJ publishes a Christmas issue full of tongue-in-cheek research that is nonetheless real science. Paramedics studying what causes patients to vomit on them. A formal investigation into whether Bond - James Bond - drinks too much. In 2011, a team of Sydney researchers asked one of the most literal questions in epidemiology: How fast does the Grim Reaper walk?¹

The conceit was that Death moves at a fixed pace, and if you can outpace it, you live. The actual question was whether a timed 6-metre walk at age 70+ could predict who would die over the next five years. The answer, it turns out, is yes - and the threshold is specific enough to double-check on a stopwatch.

The study

The data come from the Concord Health and Ageing in Men Project (CHAMP), a population-based cohort of community-dwelling men aged 70 and over recruited from Sydney’s inner-west electoral roll.¹ Voting is compulsory in Australia, so the roll is a near-complete population register - a relatively clean sampling frame. The authors note, drily, that Death was not on the electoral roll and could therefore not be recruited as a direct participant.

Participants: 1,705 men aged 70+; 49.8% born in Australia, 19.6% in Italy, the rest from Britain, Greece, China, and other countries.¹
Walking test: Trained staff used a stopwatch to time each man covering 6 metres at his usual pace. The faster of two trials was recorded, then adjusted for height.¹
Follow-up: Men were contacted by phone every four months and seen in clinic at two and five years. Deaths were verified through the NSW registry of births, deaths, and marriages.¹
Duration: Mean follow-up of 59.3 months (just under five years).
Outcome: 266 men died during follow-up. Walking speed data were available for 1,628 of the 1,705 (77 could not complete the test, mostly due to mobility limitations).

The headline numbers

The team used receiver operating characteristic (ROC) curve analysis - the same method used to calibrate medical tests - to find the walking speed that best separated who would die from who would survive.¹

Mean walking speed: 0.88 m/s (range: 0.15–1.60 m/s)¹
Youden index threshold: 0.82 m/s (~2 mph) - the speed at which sensitivity and specificity are jointly maximised, giving 63% sensitivity and 70% specificity for mortality¹
Cox regression: Men walking above 0.82 m/s were 1.23 times less likely to die (95% CI: 1.10–1.37; P = 0.0003)¹
Dose response: For every 1 m/s increase in walking speed, the hazard ratio for mortality was 2.77 (95% CI: 2.08–3.68; P < 0.001)¹
The “escape speed”: Among the 22 men (1.4%) who walked at 1.36 m/s or faster (~3 mph), none died during the follow-up period¹

Walking speeds in context

The study’s two thresholds - 0.82 m/s and 1.36 m/s - are abstract until you place them next to everyday reference points.

Axes: Each bar is a specific walking speed (horizontal axis, metres per second), placed in order from slowest to fastest. Look for: where the red and green markers sit relative to everyday benchmarks — the US crosswalk speed falls between them. Takeaway: the mortality threshold is not unusually slow. It is close to the pace street crossings are engineered for, meaning a meaningful share of older adults walk at or below it in everyday life.

What the ROC threshold actually means

A Youden index of 0.82 m/s does not mean that everyone slower than that dies - or that walking faster guarantees survival. What it means is that 0.82 m/s is the single speed at which sensitivity and specificity are jointly maximised for predicting five-year mortality in this sample.

Sensitivity 63%: Among the 266 men who died, 63% walked below the threshold. The test missed 37% of eventual decedents who happened to be faster walkers.
Specificity 70%: Among survivors, 70% walked at or above the threshold. Thirty per cent of survivors were still below it.

That is not a devastating miss rate in a biomarker. The ROC method is commonly used for laboratory tests such as cardiac troponin, and a gait-speed test performed with a stopwatch and a six-metre corridor holds up reasonably well.

Estimated 5-year mortality by speed group

Working from the sensitivity and specificity figures, it is possible to derive approximate group sizes and mortality rates.

Axes: The vertical axis is estimated five-year mortality — the percentage of each speed group who died during follow-up. The three bars are the study’s natural speed bands. Look for: the step between the slow and moderate-fast bars; the fastest group reaching zero is the far end of that trend. Takeaway: moving from the slowest group to the middle group is associated with roughly a threefold drop in mortality — an effect large enough to detect with a stopwatch in a corridor.

How to read this result

Not causal, but real: Walking speed reflects cardiovascular fitness, balance, muscle strength, and neurological integrity simultaneously. It is a composite biomarker. Telling someone to walk faster will not by itself extend their life; but the pace at which they naturally walk encodes a great deal about their overall physiological reserve.
The dose-response matters: The hazard ratio of 2.77 per 1 m/s increase is steep. Small differences in gait speed across a large population translate into substantial population-level mortality differences.
Population-specific: The CHAMP sample was community-dwelling men in inner-western Sydney - a multicultural cohort, but still men over 70 who were healthy enough to live independently and attend a clinic. The men who could not complete the walking test (n = 77) were excluded from the ROC analysis; their exclusion likely means the threshold is conservative, since the most impaired walkers were left out.
Consistent with broader literature: The 0.82 m/s threshold is nearly identical to the 0.80 m/s associated with median life expectancy across diverse populations in a 2011 JAMA meta-analysis by Studenski et al.² The two studies converged independently on almost the same number.

Practical takeaways

Self-test: A six-metre course is 20 feet - roughly the length of most hallways. Time yourself walking it at your normal pace. Divide 6 by your seconds. If the result is below 0.82, that is worth discussing with a clinician.
Track the trend: A single number matters less than whether it is declining. Annual gait-speed tracking is more informative than a one-off reading.
Treat it as a check-engine light: Declining gait speed can precede falls, hospitalisation, and cognitive impairment by months to years. Early intervention - strength training, balance work, medication review - is more effective than late intervention.
The crosswalk test is a free proxy: If you regularly catch the light on the last few seconds or need to jog across, you are close to or below the 1.22 m/s crosswalk design speed - within striking distance of the Reaper’s pace.

What the study cannot tell you

It is observational: Faster walkers almost certainly differ on dozens of unmeasured variables - genetics, diet, social engagement, absence of chronic disease. The study design cannot isolate walking speed itself as the mechanism.
It is narrow: Men, aged 70+, Sydney, one cohort. Women, younger adults, and residents of rural or car-dependent cities were not studied here. The Studenski meta-analysis suggests the threshold generalises across sexes and ethnicities, but that is a different study.
It is playful: The Christmas-paper format invites jokes. The authors lean into it - they explicitly note they could not adjust for use of the Deathly Hallows, and suggest a future “Cunning, Conniving and All-round Tricky Bastard Scale” for those who prefer to outwit Death rather than outrun it. The data and statistics underneath the jokes are real.
The 22 survivors are a small sample: The claim that 0 out of 22 fast walkers died is striking, but with only 22 people in that group, confidence intervals are wide. The “escape speed” interpretation is vivid, but it should not be taken as a hard guarantee.

The Grim Reaper walks at about 0.82 m/s in this dataset. If you can consistently clear 1.36 m/s, the data suggest you can stay ahead of him - at least for the next five years, and at least in Sydney.

The rest is up to you.

Stanaway, F. F., Gnjidic, D., Blyth, F. M., et al. (2011). How fast does the Grim Reaper walk? Receiver operating characteristics curve analysis in healthy men aged 70 and over. BMJ 343: d7679. PDF ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹
Studenski, S., Perera, S., Patel, K., et al. (2011). Gait speed and survival in older adults. JAMA 305(1): 50–58. PDF ↩

Glossary

ROC Curve: Receiver Operating Characteristic curve. A technique for finding the best threshold for a binary test — in this case, the walking speed that best separates who will die from who will survive. The curve plots sensitivity against specificity at every possible threshold; the best single point is where both are jointly maximised.
Youden Index: The point on the ROC curve where sensitivity plus specificity is highest — in other words, the threshold that does the best overall job of catching true cases while minimising false alarms. The 0.82 m/s figure is the Youden-index threshold for five-year mortality in this study.
Sensitivity: The proportion of people who actually experienced the outcome (death) that the test correctly identified as high-risk. A sensitivity of 63% means the test caught 63% of eventual deaths — it missed the 37% who happened to walk faster than the threshold.
Specificity: The proportion of people who did not experience the outcome (survivors) that the test correctly identified as low-risk. A specificity of 70% means 70% of survivors walked at or above the threshold — 30% of survivors were still in the 'slow' band.
Cox Regression: A standard statistical technique for survival data that estimates how strongly a variable (here, walking speed) is associated with the time until an event (death), after accounting for random chance. It produces a hazard ratio.
Hazard Ratio(HR): A measure of how much more (or less) likely an event is in one group versus another over time. A hazard ratio of 2.77 per 1 m/s increase in walking speed means each extra metre per second of pace was associated with 2.77 times the chance of surviving the follow-up period.
Composite Biomarker: A single measurable quantity that reflects several underlying health factors at once. Walking speed is a composite biomarker because it simultaneously encodes cardiovascular fitness, muscle strength, balance, and neurological integrity — none of which are measured directly.