The CDC's National Center for Health Statistics put the 2025 US age-adjusted death rate at 689.2 per 100,000, down 4.6% from 722.1 in 2024 and the lowest figure on record (CDC/NCHS, July 2026). For life, annuity, and pension actuaries, the number that matters is not the record itself but that it is the second consecutive annual decline, arriving on top of a scale most reserving models have not moved.

The Vital Statistics Rapid Release Report No. 44, based on 99.9% of 2025 death records processed as of May 10, 2026, counted 3,094,593 total US deaths for the year, with rates falling for every age group and for both sexes (CDC/NCHS, July 2026). Farida Ahmad, a CDC health scientist and co-author of the report, credited a continued decline in fatal drug overdoses as a primary driver of the low. That single mechanism, an overdose count still working its way down from a 2021-2022 peak, is doing a disproportionate share of the work in a statistic that population actuaries, pension funds, and annuity writers are about to spend the next reserving cycle arguing over.

The Record, Decomposed

Men died at 811.1 per 100,000 in 2025 against 582.9 for women, a 228.2-point gap that has narrowed only slightly from the 2024 comparison and remains the widest sex differential among the major mortality inputs actuaries track (CDC/NCHS, July 2026). Heart disease stayed the leading cause of death at 694,708, followed by cancer at 622,832 and unintentional injuries at 184,265, while influenza and pneumonia climbed from the eleventh leading cause to the eighth with 56,511 deaths, a shift the report links to the same reporting-lag dynamics that make every provisional release subject to later revision (CDC/NCHS, July 2026).

Cause of death2025 deaths (provisional)Rank
Heart disease694,7081
Cancer622,8322
Unintentional injuries184,2653
Influenza and pneumonia56,5118 (up from 11)

The overdose component is the clearest driver behind the mechanism. An estimated 70,000 Americans died of drug overdoses in 2025, a 14% drop from 2024 and the third consecutive annual decline, with fentanyl, cocaine, and methamphetamine deaths all falling even as Arizona, Colorado, and New Mexico bucked the national trend (CDC/NCHS, July 2026). That third straight year of overdose decline follows two years in which the same trend already pulled overall US life expectancy up to 79.0 years in 2024, an increase of 0.6 year from 78.4 in 2023, with the male-female life expectancy gap narrowing to 4.9 years as male life expectancy gained more ground, 76.5 years versus 81.4 for women (NCHS Data Brief 548, January 2026). The 2024 age-adjusted death rate itself fell 3.8% from 750.5 in 2023 to 722.1, meaning the population has now logged two back-to-back annual improvements of comparable magnitude, not a single post-pandemic rebound year followed by a plateau.

Two Years Running: What Resumed Improvement Means for the MP-2021 Debate

That two-year pattern lands directly on an open question in actuarial mortality-improvement modeling. actuary.info's coverage of the SOA's 2026 Mortality Improvement Model reported that the Retirement Plan Experience Committee extended the underlying dataset through 2023 but left the MP-2021 projection scale itself unchanged, concluding there was insufficient post-pandemic data to justify a new scale even as pension risk transfer deals were pricing at 99.7% of accumulated benefit obligation. The 2025 CDC release adds a data point the committee did not have when it made that call: a second consecutive year in which the general population improved faster than the frozen scale assumes, following the 2023-to-2024 improvement it did have.

The mechanism is not abstract. IRS Notice 2026-27, published in Internal Revenue Bulletin 2026-21 in May 2026, sets the mortality tables qualified defined benefit plans must use for 2027 minimum funding and lump sum calculations under IRC section 430(h)(3)(A), and that framework, described in actuary.info's analysis of the 2027 DB mortality tables, builds from the Pri-2012 base tables, applies the RPEC MP-2021 improvement scale, caps annual improvement factors at 0.78% under the SECURE 2.0 Act, and holds a zero-improvement window for 2020 through 2023 to reflect COVID-era excess mortality. That zero-improvement window was a reasonable response to a pandemic that killed disproportionately among older adults. It was not designed to persist once population mortality resumed improving at a pace exceeding the capped 0.78% assumption two years running, which is what the CDC's 2024 and 2025 releases together now show. A capped, frozen improvement scale sitting downstream of a population trend that has outrun it in consecutive years is a mechanical mismatch, not a modeling choice, and it is the kind of mismatch that widens with every additional year the scale goes unrevised.

The population figure and an actuarial improvement scale are not strictly the same measurement, and the distinction matters for how much weight to place on the CDC number directly. VSRR No. 44 reports a crude, age-adjusted death rate across the entire population, a blend of true cohort-level mortality decline and compositional shifts, most visibly the drop in overdose deaths concentrated among ages 25 to 54, that changes which age groups are dying at what rate rather than uniformly slowing mortality at every age. MP-2021, by contrast, projects age-and-sex-specific improvement factors meant to isolate the cohort effect from that kind of cause-mix noise. Even accounting for the distinction, the scale of the mismatch is hard to dismiss: a 4.6% single-year drop in the age-adjusted rate is roughly six times the 0.78% annual improvement IRS Notice 2026-27 permits the pension mortality basis to reflect, before compounding a second comparable decline from 2024. No actuary should treat a single population statistic as a ready-made replacement improvement factor. But a capped scale sitting an order of magnitude below what the population has delivered for two years running is the kind of gap that eventually surfaces in an experience study rather than staying a modeling footnote.

The Annuitant and Pensioner Reserve Math

For annuity and pension actuaries, the direction of that mismatch cuts one way: understated future improvement means understated liabilities. A single-premium immediate annuity or a defined benefit pension promises a level payment stream for as long as the annuitant lives, so any improvement scale that assumes slower mortality gains than the population is actually delivering will project the liability's expected duration too short and its present value too low. The 2020-2024 Individual Payout Annuity Mortality Experience Study covered separately on this site found the same tension in insured payout annuitant data specifically, the first industry benchmark to run the entire COVID-19 mortality arc, arriving as SPIA sales climbed and forcing the question of whether surviving annuitants now live longer than the 2012 IAM Table assumes. The CDC's population-level confirmation of a second straight improvement year strengthens that case rather than resolving it: an insured annuitant population is healthier and wealthier than the general population by underwriting selection, so if the general population is improving faster than MP-2021 assumes, the annuitant subset the improvement scale is actually meant to project has every reason to be improving at least as fast.

Pension plan sponsors are watching the same signal through a different lens. Corporate defined benefit plans that locked in favorable funded status through 2026, a trend covered in actuary.info's reporting on Q2 2026 pension surplus positions, priced their endgame and settlement strategies against the same frozen mortality assumptions now facing two years of contrary population evidence. A plan actuary setting the discount rate and mortality basis for a 2027 valuation has to decide whether to hold the regulatory-minimum IRS table, which will not move until the next SECURE 2.0 review cycle, or layer in a company-specific improvement adjustment that reflects the CDC's confirmed trend, a choice with direct funded-status and PBGC premium consequences given that flat-rate premiums are now assessed per participant regardless of the underlying mortality basis used to compute the liability itself.

Some sponsors are sidestepping the assumption fight rather than resolving it. A longevity swap, structured so the plan pays a fixed schedule of expected benefit payments and receives actual payments as they emerge, moves the entire question of whose improvement assumption is correct onto a reinsurer's balance sheet, an alternative to a full buyout that actuary.info has covered separately as a middle path between retaining plan assets and eliminating demographic uncertainty outright. A reinsurer pricing a new swap in 2026 has access to the same two years of CDC data as the plan sponsor's own actuary, so the population signal does not resolve the underlying pricing gap so much as relocate it to whichever counterparty's mortality model updates first.

Where the Improvement Doesn't Help: Term Life and the Under-40 Overdose Story

The same record death rate that pressures annuity and pension reserves works against life insurers in the opposite direction, and not uniformly. actuary.info's earlier analysis of the SOA's 2025 Individual Life Insurance Mortality Improvement Scale found the mortality picture "mostly, but not entirely, normalized," with a small residual of excess mortality still visible among ages 65 and older even as the broader population improved. A lower population death rate compresses the margin between a term life carrier's priced mortality assumption and emerging experience only where the carrier's pricing had not already assumed continued improvement, and pricing actuaries who built in aggressive improvement assumptions during the 2023-2024 rebound now face emerging experience that, while favorable in absolute terms, may already be within the range their pricing anticipated rather than a windfall above it.

The overdose component carries specific relevance for underwriting ages below 40, where a SOA research report on drug overdose mortality trends found that overdose deaths account for roughly 25% to 30% of all deaths among ages 25 to 39, with the average age at overdose death running about 30 years younger than all-cause mortality generally (SOA, 2025). Proprietary claims data from RGA cited in that same report showed insured-population overdose claims peaking in 2021 and declining since, mirroring the general population pattern, with elevated mortality concentrated among smokers, policies under $100,000 in face value, and later policy durations. A term life book concentrated in that age-and-face-amount profile has more genuine mortality-improvement upside from the overdose decline than a book weighted toward older, larger-face permanent business, where cardiovascular and cancer mortality, not overdose, set the pace, and both causes of death fell more modestly in the CDC's 2025 release than the overdose-specific number did.

The Longevity-Risk-Transfer Read

Pension risk transfer pricing sits at the intersection of both effects, and the market has been volatile enough this year to make the mortality question harder to isolate from timing noise. LIMRA's U.S. Group Annuity Risk Transfer Sales Survey recorded single-premium PRT sales jumping 132% in the fourth quarter of 2025 to $28 billion, only for first-quarter 2026 sales to fall 47% year over year to roughly $3.8 billion across 102 contracts, a pattern LIMRA attributed to deal-flow timing rather than a change in underlying sponsor demand. A record-low population death rate arriving in the middle of that swing raises the stakes on how PRT insurers set the longevity basis for pricing new buyout and buy-in blocks: a carrier that prices a 2027 PRT transaction off a mortality table that has not absorbed two years of confirmed population improvement is underpricing the longevity risk it is assuming, a miscalculation that only shows up years later as the block runs longer than reserved.

Regulators are moving on a parallel track. actuary.info's coverage of the NAIC's C-2 Longevity Risk RBC framework reported that the Academy of Actuaries' Longevity Risk Task Force and the NAIC's Longevity Risk Subgroup reached agreement in spring 2026 on the direction of a scenario-based longevity stress charge for retained and ceded longevity exposure, targeting a year-end 2027 implementation. A scenario-based capital charge is, by design, meant to capture exactly the kind of tail outcome where realized population mortality improvement outpaces the pricing assumption for several years running, which is the scenario the CDC's 2024-2025 data is now actively building rather than merely illustrating. The C-2 framework's field test protocol, expected once the Subgroup finalizes its response, will be the first real opportunity to see whether regulators plan to calibrate that stress against the improvement pattern the population data is currently showing or against the older, more conservative baseline still embedded in MP-2021.

Further Reading

Sources