What are the odds of living through a prolonged power-grid blackout lasting days or longer?
Evidence quality 4.63/5
Eight-dimension review score against the quality rubric . Each dimension scored 1–5.
- D1 Source grounding
- 5/5
- D2 Source authority
- 5/5
- D3 Arithmetic
- 4/5
- D4 Uncertainty
- 4/5
- D5 Scope
- 5/5
- D6 Prose
- 5/5
- D7 Perception honesty
- 4/5
- D8 Caveat completeness
- 5/5
● your factors — click this risk ▾ to reveal
- Your factors
≈ As likely as
Perceived
Public worry about a "grid-down" event is loud but oddly bimodal. One mode is the prepper-and-thriller framing — a cyberattack or solar storm that plunges a continent into darkness for weeks — which most people rate as remote and cinematic. The other mode is the mundane multi-day outage after a hurricane, ice storm, or derecho, which most people treat as a weather nuisance rather than a "blackout" at all and therefore underweight as a lifetime probability. The result is a perception focused on the dramatic cause (will hackers take down the grid?) while the far more likely path — a regional weather-driven outage lasting one to several days — is mentally filed under routine inconvenience. No major survey isolates worry about a multi-day outage from any cause; the closest proxies (Chapman's cyberterrorism and power-grid-failure items) measure the catastrophic framing, not the ordinary one.
Rough estimate: Most adults rate a multi-day outage as a weather nuisance, not a lifetime probability; the catastrophic 'grid-down' framing is what draws worry
Source: editorial intuition, not polled
Actual
~5% per year that a US household experiences a power outage lasting 24 hours or longer
US households, multi-day (≥24h) electricity outage
Show derivation
This entry isolates the cause-agnostic exposure question — the probability that a typical US adult personally lives through at least one prolonged (≥24-72 hour, up to weeks) regional grid outage over an adult lifetime, from ANY cause: severe weather, equipment failure or cascading fault, cyberattack, or solar storm. It deliberately does not condition on the cause, which separates it from cyberattack-infrastructure and solar-storm-carrington-event, where a blackout is the consequence of one specific trigger. Step 1, native annual rate: the 2023 American Housing Survey (Census/HUD) found 1 in 4 US households (33.9 million) had at least one complete outage in the prior 12 months, and ~70% of those (23.6 million, or ~18% of all households) had an outage lasting 6 hours or longer. Narrowing from the 6-hour threshold to the ≥24-hour threshold the question fixes: multi-day outages are the weather tail of that distribution. Do et al. (Nature Communications 2023) found 62.1% of 8+ hour outages co-occur with extreme weather and that 70.5% of US counties saw at least one 8+ hour outage over 2018-2020, with a county median of 2 per year. Taking roughly a quarter to a third of the ~18% of households with a 6+ hour outage as reaching ≥24h gives a central native rate of ~5% per household-year (1 in 20). Step 2, lifetime: the per- person annual rate is strongly location-dominated and autocorrelated — your grid, your weather region, and the fact that adults rarely relocate set the rate — so naive compounding of a population-average annual rate over 59 remaining adult years (1-(1-0.05)^59 ≈ 0.95) overstates the population-average lifetime probability by Jensen's inequality. Instead the lifetime figure is bracketed two ways. Lower anchor, catastrophic regional events only: the 2003 Northeast blackout reached ~50 million customers (DOE) and Winter Storm Uri in 2021 shed 23,418 MW (FERC/NERC, the largest firm load shed in US history) affecting millions for up to four days; major regional multi-day events of that class recur roughly every few years and each exposes tens of millions, giving a lifetime exposure to a headline-scale event of ~0.3-0.4. Upper anchor, all weather-driven multi-day local outages (residents of hurricane, ice-storm, and derecho regions accumulate these across decades): ~0.6-0.85. The central estimate of ~0.5 (1 in 2) sits between these, with a band reflecting the heterogeneity between grid-reliable low-risk areas and the hurricane/ice belt.
Caveats: The headline ~1 in 2 lifetime figure is the probability of personally experienci…
The headline ~1 in 2 lifetime figure is the probability of personally experiencing the outage, not of being harmed by it. Most multi-day outages are an expensive inconvenience: spoiled food, no heat or cooling, lost work, and a few days of disruption. The mortality tail is real but far smaller — Winter Storm Uri was associated with at least 246 deaths in Texas (and some estimates run higher), concentrated among the elderly, the medically power-dependent, and households without alternate heat. That tail, not the typical event, is what the death-from-heat or death-from-cold framings would capture. Two structural caveats apply to the number itself. First, it is deliberately cause-agnostic: weather drives the overwhelming majority of multi-day outages (Do et al.: 62.1% of 8+ hour outages co-occur with extreme weather; EIA: 80% of 2024 outage-hours from hurricanes), while cyberattack and solar storm are low-frequency tails included for completeness — the Lloyd's/Cambridge Erebos scenario shows the cyber tail is reachable but it is a scenario, not a frequency. Second, the lifetime figure is bracket-derived rather than compounded: multi-day outage risk is dominated by where you live and is highly autocorrelated, so the population average masks a wide spread — a hurricane-belt rural customer may be near-certain to see several multi-day outages, while a customer on a dense undergrounded urban network may go a lifetime without one outside a catastrophic regional event. The ~5% native annual rate is itself an estimate that narrows the AHS 6+ hour figure to the ≥24h threshold using the weather-tail structure in Do et al.; it is the largest single source of uncertainty in the entry.
Related risks
Other risks on similar themes — for exploring related fears.
Career obsolescence
What are the odds of needing to change careers due to technological disruption?
Pick challenger
The cleanest base rate for this question is a household survey, not a grid-engineering report. The 2023 American Housing Survey found that 1 in 4 US households were completely without power at least once in the prior year, and roughly 70% of those — about 18% of all households — had an outage lasting six hours or more. Narrowing to the question’s threshold, a prolonged outage of a day or longer, means looking at the weather tail of that distribution: Do et al. (Nature Communications, 2023) found that 62.1% of 8+ hour outages coincide with extreme weather, and that 70.5% of US counties experienced at least one such outage in just three years (2018-2020). Combining the household incidence with the multi-day fraction puts the native rate at roughly 5% per household-year for an outage lasting 24 hours or more. Over an adult lifetime, that points to a probability near 1 in 2.
The arithmetic deliberately avoids naive compounding. A 5% annual rate compounded over 59 remaining adult years would give about 95%, but that figure is wrong for a structural reason: multi-day outage risk is dominated by where you live, and people rarely move far. The annual rate is highly autocorrelated, so compounding a population average overstates the population-average lifetime probability. A bracket is more honest. On the low side, counting only catastrophic regional events — the 2003 Northeast blackout reached ~50 million customers (US Department of Energy), and Winter Storm Uri in 2021 produced 23,418 MW of firm load shed, the largest in US history (FERC/NERC), leaving millions without power for up to four days — gives a lifetime exposure of roughly 0.3 to 0.4. On the high side, counting all weather-driven multi-day local outages that hurricane, ice-storm, and derecho-belt residents accumulate over decades pushes toward 0.6 to 0.85. The midpoint, about 1 in 2, is the headline; the wide band is the genuine spread between a reliable urban grid and the Gulf coast.
Where the number does not apply is the more interesting part. This entry is cause-agnostic on purpose, and that reframes the fear. The cinematic causes — a cyberattack like the Lloyd’s/Cambridge Erebos scenario that leaves 93 million people across 15 states dark, or a Carrington-class solar storm — are real tails but rare ones; weather drives the overwhelming majority of multi-day outages, and EIA data shows the weather share is rising (hurricanes Beryl, Helene, and Milton accounted for 80% of all US outage-hours in 2024, when the national average doubled to 11 hours). So the honest framing is calibrated, not overrated: the grid-collapse scenario most people picture is unlikely, but the ordinary version of the same event — sitting in the dark for two or three days after a storm — is something close to a coin-flip over a full adult life, and closer to a near-certainty if you live where the weather is. The figure measures experiencing the outage, not surviving it; the small mortality tail, concentrated among the elderly and the medically power-dependent, is the part the headline number does not carry.
Related tidbits
About 1 in 4 US households reports a 6+ hour power outage in a single year, and over a lifetime, living through a multi-day regional blackout is close to a coin flip. Roughly 62% of the long outages track extreme weather — and the rate is climbing.
Claim ledger
Every number below is what each source reported, with the verbatim quote we relied on and how we arrived at our figure. Click any link to verify directly.
-
[1] U.S. Census Bureau (2023 American Housing Survey, sponsored by HUD) — About 1 in 4 Households Experienced a Power Outage in the Span of a Year
About 1 in 4 Households Experienced a Power Outage in the Span of a Year- Statistic
1 in 4 US households (33.9 million) had a complete power outage in the prior 12 months; ~70% (23.6 million) had an outage lasting 6 hours or more- Excerpt
“"About 33.9 million or 1 in 4 households nationwide reported they were completely without power at least once in the 12 months before they were interviewed for the 2023 American Housing Survey (AHS). … Approximately 70% or 23.6 million of the households reporting an outage said at least one outage lasted 6 hours or more." ”
- Source data from
- 2024-10-02
- Accessed
- 2026-06-21 · archived copy
- Calculation
- The AHS is the cleanest household-level base rate for outage exposure: nationally representative, self-reported, and duration-bucketed. The publicly summarized figure tops out at the 6+ hour category (~18% of all households per year). This entry narrows from that 6-hour threshold to the ≥24-hour threshold the question fixes by treating multi-day outages as the long weather tail of the 6+ hour distribution (see Do et al.), estimating roughly a quarter to a third of 6+ hour household-outages reach ≥24h, i.e. ~5% per household-year. The AHS sets the upper envelope: if 18% of households see a 6+ hour outage in a single year, multi-day exposure over a 59-year adult life is common, not rare.
- Independence
- Household survey methodology (self-report, nationally representative sample), independent of the utility-reported SAIDI data and the county-level outage-detection methodology in Do et al.
-
[2] Nature Communications / Do V, McBrien H, Flores NM, et al. — Spatiotemporal distribution of power outages with climate events and social vulnerability in the USA
Spatiotemporal distribution of power outages with climate events and social vulnerability in the USA- Statistic
62.1% of 8+ hour outages co-occur with extreme weather/climate events; 70.5% of studied US counties (73.7% of population) had at least one 8+ hour outage over 2018-2020, county median 2 per year- Excerpt
“"62.1% of 8+ hour outages co-occur with extreme weather/climate events, particularly heavy precipitation, anomalous heat, and tropical cyclones. … 8+ hour outages were 3.4x more common on days with a single event and 10x more common on days with multiple events." ”
- Source data from
- 2023-04-29
- Accessed
- 2026-06-21 · archived copy
- Calculation
- Do et al. analyzed ~231,000 1+ hour outages and ~17,500 8+ hour outages across 2,447 counties (73.7% of US population) from 2018-2020 using utility outage-tracker data. The key contribution for this entry is the bridge from "any 6+ hour outage" to "prolonged multi-day outage": 8+ hour outages are weather-driven (62.1% co-occur with extreme weather) and the multi-day tail is the hurricane/ice/derecho tail of that same distribution. The finding that 70.5% of counties saw at least one 8+ hour outage in just three years independently corroborates that long-duration outage exposure is common over a lifetime, supporting a lifetime estimate near 1 in 2 rather than the low single-digit annual rate alone would naively suggest.
- Independence
- Uses county-level utility outage-tracker detection, methodologically independent of the AHS self-report survey and the FERC/NERC and DOE event reconstructions.
-
[3] U.S. Energy Information Administration — Hurricanes in 2024 led to the most hours without power in the United States in 10 years
Hurricanes in 2024 led to the most hours without power in the United States in 10 years- Statistic
US customers averaged 11 hours of electricity interruptions in 2024 (~2x the prior decade's average); major events accounted for 80% of those hours; major-event interruptions averaged ~9 hours vs ~4 hours/year in 2014-2023- Excerpt
“"U.S. electricity customers experienced an average of 11 hours of electricity interruptions in 2024, or nearly twice as many as the annual average experienced in the decade before. … Major events such as Hurricanes Beryl, Helene, and Milton accounted for 80% of the hours without electricity in 2024." ”
- Source data from
- 2025-12-01
- Accessed
- 2026-06-21 · archived copy
- Calculation
- EIA's SAIDI data establishes two things this entry relies on. First, the trend: average interruption hours doubled from ~5.6 (2022) to 11 (2024), driven almost entirely by major weather events (80% of 2024 hours), which is the era/climate factor behind the rising-trend multiplier. Second, the structure: routine non-major interruptions hold steady at ~2 hours/year, so the multi-day outages this entry counts are overwhelmingly the major-event tail, not everyday flickers. SAIDI is an average-hours metric, not a per-household incidence rate, so it is used here for trend and attribution rather than as the native probability — that comes from the AHS.
- Independence
- Utility-reported SAIDI/SAIFI aggregated by EIA, independent of the AHS household survey and the academic county-level analysis.
-
[4] Lloyd's of London / University of Cambridge Centre for Risk Studies — Business Blackout: The insurance implications of a cyber attack on the US power grid
Business Blackout: The insurance implications of a cyber attack on the US power grid- Statistic
Hypothetical Erebos cyberattack on 50 Northeastern US generators leaves 93 million people across 15 states (including NYC and Washington DC) without power; ~$243bn to >$1trn US economic impact- Excerpt
“"The attackers are able to inflict physical damage on 50 generators which supply power to the electrical grid in the Northeastern USA, including New York City and Washington DC. … it triggers a wider blackout which leaves 93 million people without power. … The total impact to the US economy is estimated at $243bn, rising to more than $1trn in the most extreme version of the scenario." ”
- Source data from
- 2015-07-06
- Accessed
- 2026-06-21 · archived copy
- Calculation
- This source supplies the cause-agnostic tail: a cyberattack is one of the four causes this entry folds together, and the Lloyd's/Cambridge Erebos scenario quantifies the worst-case cyber path (93 million people, 15 states). It is a scenario study, not a frequency estimate, so it does not enter the probability arithmetic — it bounds the severity tail and demonstrates that the multi-day regional blackout this entry counts is reachable by cyber means, not only weather. No publicly fetched figure supports a "1:200 return period" attribution, so that claim is deliberately not cited here.
- Independence
- Insurance-industry scenario modelling by the Cambridge Centre for Risk Studies, independent of all three empirical outage datasets above.