FAA CARI-7 explained: how aircrew cosmic-radiation dose is actually calculated

When a newspaper reports that a New-York-to-London round trip delivers about 0.1 millisievert of cosmic radiation, that number almost certainly came, directly or via a derivative, from a run of the FAA's CARI series. The current version, CARI-7, with its web-accessible front end CARI-7A, replaced CARI-6 in 2017 and is the operational dosimetry reference for aircrew in the United States ^[1]. The model is freely available, freely runnable, and well documented in a series of DOT/FAA/AM technical reports authored by Kyle Copeland and colleagues at FAA CAMI in Oklahoma City ^[2].

What follows is CARI-7 in operational terms: what it computes, what inputs it needs, how it differs from CARI-6, and where its ±25% uncertainty comes from. Primary FAA and ICRP references are linked as they come up. This is not a replacement for the CAMI user manuals; for those, DOT/FAA/AM-17/4 and its successors are indexed from the sources page.

What CARI-7 actually is

CARI-7 is, in the simplest terms, a particle-transport calculator. It models the cascade of secondary particles produced when galactic cosmic rays (GCR), overwhelmingly protons and helium nuclei from outside the solar system, strike the top of Earth's atmosphere, propagate downward, and produce neutrons, muons, pions, electrons, and photons that deliver dose to a body cruising in the stratosphere. CARI-7 integrates that complex radiation field along the flight path of a specified aircraft and reports the time-integrated effective dose in microsieverts.

Effective dose is an ICRP-defined quantity that weights absorbed dose by the biological effectiveness of the radiation type (the radiation weighting factor w_R) and by the differential radiosensitivity of organs (the tissue weighting factor w_T). CARI-7 uses the w_T values from ICRP Publication 103 (2007) ^[3]. The 2007 weighting set replaced the 1990 ICRP-60 set; FAA CAMI's transition to ICRP-103-weighted output is one of the headline differences from CARI-6.

Inside the model: what changed from CARI-6 to CARI-7

CARI-6 used parameterised tables of dose rate as a function of altitude, geographic location, and date, derived from earlier transport calculations and tuned against in-flight measurements. It was fast (a single-segment lookup) but its physics fidelity was limited by the underlying tabulation. CARI-6 was the operational reference from 1997 through the 2010s.

CARI-7 replaced the parameterised tables with a full Monte Carlo transport calculation built on MCNPX, the Monte Carlo N-Particle eXtended code maintained by Los Alamos National Laboratory. The relevant CARI-7 technical reports describe how Copeland and CAMI generated a dense lookup grid of MCNPX runs across altitude, vertical-cutoff rigidity (a geomagnetic-shielding proxy), and heliocentric potential (a solar-cycle proxy), and how CARI-7 interpolates among those runs for any given flight ^[2]. Three concrete consequences:

• Better treatment of secondary neutrons. Neutrons account for roughly half of effective dose at cruise altitudes, and the secondary neutron field is heavily energy-dependent. MCNPX models the full secondary spectrum explicitly; CARI-6's parameterisation did not.
• Better treatment of high-altitude / high-latitude regimes. Polar high-altitude flying is exactly where the CARI-6 tabulation had the most extrapolation; CARI-7's grid covers that region with native MCNPX runs.
• ICRP-103 weighting throughout. CARI-7 reports E using ICRP-103 w_T; CARI-6 output had to be reinterpreted by hand for users who needed ICRP-103 numbers.

The inputs CARI-7A accepts

The public web tool, CARI-7A, asks for five things ^[1]:

1. Origin airport. IATA or ICAO code. Maps internally to latitude/longitude and elevation.
2. Destination airport. Same.
3. Cruise altitude. Entered as flight level (e.g. FL370) or altitude in feet or metres. CARI-7A also accepts a stepped profile if you want to model an early step-climb.
4. Departure date. Used to retrieve the heliocentric potential, a parameter that captures how strongly the heliospheric magnetic field is suppressing GCR flux at that point in the 11-year solar cycle. During solar maximum, heliocentric potential is high and GCR flux at Earth is suppressed; during solar minimum, the field is weak and GCR flux rises. CARI's monthly heliocentric-potential table is maintained by FAA CAMI from neutron-monitor data (primarily Oulu in Finland).
5. Climb / descent profile (optional). CARI-7A assumes default climb and descent rates for civil airliners if you don't specify; you can override for a specific operation.

What CARI-7A does not ask for:

• Aircraft type. The model uses a generic civil airframe with default shielding mass. Differences between, say, an Airbus A220 and a Boeing 777 at the same flight level are within CARI's stated uncertainty.
• Cabin position. The model produces a single dose for the aircraft interior; it does not differentiate seat row or column.
• Passenger characteristics. The output is effective dose to an ICRP reference person; neither sex nor age nor pregnancy status changes the CARI output. Those factors enter at the interpretation stage, not the model stage.

The outputs CARI-7A produces

The principal output is a single number: the per-flight effective dose in microsieverts, with ICRP-103 weighting. For a typical New York (JFK) to London (LHR) cruise at FL360 in mid-2026, CARI-7A returns approximately 50 µSv per one-way segment, depending on the exact great-circle routing and the solar-cycle phase. A polar transpolar (e.g. New York to Hong Kong over the pole at FL400) can return 90 to 120 µSv per one-way segment.

Behind the single number, the model also reports the breakdown by radiation type: neutron, proton, photon-and-electron-cascade, muon. For typical cruise conditions, neutrons supply roughly 45–55% of effective dose, the electromagnetic cascade roughly 25–35%, and protons / muons the remainder. The exact split varies with cutoff rigidity and altitude.

How CARI-7 was validated

CARI's validation history is the reason it is the operational reference and not a competing code. The CARI-series papers by Friedberg, Copeland, and colleagues document comparisons of CARI output to in-flight measurements made with tissue-equivalent proportional counters (TEPCs), Bonner-sphere neutron spectrometers, and silicon detector packages across thousands of commercial-flight legs over four decades ^[4]. Independent measurements by European groups (EURADOS) and by carriers operating their own dosimetry programs under EURATOM directives have largely corroborated CARI's predictions within roughly ±25% at the 2σ level for galactic-cosmic-ray-driven dose at typical cruise altitudes.

That ±25% is the headline uncertainty you should carry mentally for any CARI-derived dose number. It is not a flaw of the code; it reflects the inherent variability of the cosmic-ray field, the difficulty of in-flight measurement, and the fact that effective dose is itself a model-defined quantity, not a directly measurable one. For more detail see our methodology page.

What CARI-7 deliberately does not do

Outside CARI-7's scope	Reason
Solar particle event (SPE) dose	The SPE spectrum changes event by event and CARI's monthly heliocentric-potential model cannot capture the impulsive spectral shape. SPE dose has to be modelled per-event with codes like NAIRAS or the FAA's separate SPE response procedure. See our SPE guide.
Suborbital or orbital dose	CARI's atmospheric depth model breaks down above the stratosphere. For X-15-class, suborbital tourism, or LEO, codes like HZETRN are the correct tool.
Aircraft-specific shielding differences	Within civil airliners these are below the model's uncertainty band; for unusual airframes (military, experimental) CARI's generic shielding assumption is not appropriate.
Individual cancer risk	CARI reports effective dose. Translating dose to individual risk requires BEIR VII / ICRP-103 risk coefficients and assumptions about the dose-response relationship that are well outside the model.

Why FlightRadiation runs CARI per segment

The free CARI-7A web tool is excellent for single-segment work. The reason we built FlightRadiation is that almost no real flier has a single segment; they have a year of segments. Annualisation, polar-route attribution, and ICRP-limit comparison all require running CARI per segment, summing, and presenting against the right reference frames. The underlying physics is CARI; the value FlightRadiation adds is the framing.

Our methodology page documents exactly how we use the CARI tables, our heliocentric-potential refresh cadence, and the routing model we use for the polar-attribution metric. Where there is any divergence between our implementation and the published CARI-7A web output for the same inputs, we treat it as a bug and fix it.

How to read a CARI-7 number

Given the above, a useful mental decoder for any CARI-derived dose value:

• It is effective dose, ICRP-103 weighted, not absorbed dose or equivalent dose to a single organ.
• It applies to an ICRP reference person, not to you individually.
• It has an uncertainty of roughly ±25% at 2σ for GCR-driven dose at typical cruise altitudes.
• It does not include SPE dose unless explicitly modelled separately.
• It depends on solar-cycle phase via the heliocentric potential, so re-running an old flight with today's solar-cycle phase will give a slightly different answer.

With those five caveats internalised, CARI-7 numbers are extraordinarily useful for decision-making. The model is mature, conservative, and freely available. We use it because there is no better tool, and we present it within ICRP-103 reference limits because numbers in millisieverts mean nothing without context. Read on for the rest of the series.

Last reviewed 30 June 2026 · See our methodology and sources.