Research

The research behind Sunwize

Every number Sunwize uses to count your sun, why we chose it, and where it falls short. When the app changes, this page changes with it.

Updated 13 November 2026Three research reviews, 66 sourcesOur reading of published studies

At a glance

  • Your limit is first faint redness, not a safe dose. 100% is our estimate of the UV dose at which a typical person with your skin type first shows faint pink the next day.
  • Six limits, each on a measured or official figure: 2, 2.5, 3.5, 5, 7.5 and 12 SED for skin types I to VI.
  • We count the full dose a flat surface gets. Most of your skin gets less, but bare shoulders get all of it.
  • The day adds up. Going inside pauses the count. It doesn’t reset it. A new day starts at midnight.
  • Sunscreen counts at a third of its SPF, for 2 hours, because people put on far less than the lab does.
  • A heads-up at 75% of your limit, or 15 minutes before it, whichever comes first.

What 100% means

100% is Sunwize’s estimate of the sunlight dose at which a typical person with your skin type first shows faint redness the next day, on unprotected, untanned skin that faced the sun the whole time, such as bare shoulders.

Many people redden sooner and some later. It’s about half the dose of a red, tender sunburn. It isn’t a safe amount of sun: skin damage starts well below it.

Your limit, 100%Typical first faint pink, the next day

Heads-up75%

Skin damage already measurableDNA damage from about 20%, immune effects from 25–50%

Red, tender sunburnAbout 2 to 3 times the limit

Dose as a share of your limit. Damage: Shih 2018, in all six skin types. Sunburn: 2 to 3 times the first-pink dose “induced a vivid but nonblistering sunburn in all volunteers” (Kelly 2000, Harrison 2004).

Why this definition

  • Every limit is compared with the same lab measure, the minimal erythema dose (MED): the smallest dose that leaves just-visible pink about 24 hours later.
  • It’s the only definition the evidence supports for all six limits. None of them can be defended as a dose below which you won’t burn or won’t be harmed.
  • It explains why people often pass 100% with no burn they’d notice. A sunburn people recognise takes 2 to 3 times the first-pink dose. In a Danish diary study, people’s dose on days they reported sunburn was 2.5 times their dose on comparable days without one (Thieden 2005). And most skin gets only part of the counted dose (see how it’s counted).

The dose is measured in SED, the standard unit of sunburning UV: 1 SED is 100 J/m² of UV, weighted by how strongly each wavelength reddens skin.

Your limit by skin type

Each limit sits on the average first-redness dose measured in phototest studies, or on a figure published by health authorities and standards bodies. Open a skin type to see the case for its number.

Type IBurns very easily2SED13 min at UV 10Strong evidence

Evidence base: 4 study samples, 27 people, and five official figures.

Why 2 SED

  • Four phototest studies put the type I average at 2.1 to 2.3 SED (Shih 2018, Shih 2020, Pérez Ferriols 2014, Valbuena 2020).
  • ARPANSA: “A very fair-skinned person can receive sunburn and skin damage from only two SEDs” (ARPANSA). Health Canada, German sunbed law and Japan’s NIES use the same 2 SED.
  • It matches the 11 minutes at UV Index 12 that ARPANSA scientists give for type I (Gies 2002).

What argues against it

  • About 4 in 10 people with type I skin would show first pink before 2 SED. Individuals were measured as low as 1.0 SED.
Type IIBurns easily2.5SED17 min at UV 10Strong evidence

Evidence base: 6 study averages and every official point value.

Why 2.5 SED

What argues against it

  • One study measured 3.7 SED in ten people (Felton 2016), and ICNIRP’s indicative range runs up to 3. The same lab measured 2.6 SED in other volunteers.
  • About half of type II would still redden before 2.5 SED, because it sits on the average.
Type IIISometimes burns3.5SED23 min at UV 10Strong evidence

Evidence base: 6 study averages and three official figures.

Why 3.5 SED

What argues against it

  • One study measured 4.2 SED in 19 people (Shih 2020); the same group measured 3.2 SED two years earlier.
  • Japan’s NIES uses 3 SED, and Danish guidance converts to about 2.7.
Type IVBurns a little5SED33 min at UV 10Good evidence

Evidence base: 4 study samples, 63 people.

Why 5 SED

What argues against it

  • German sunbed law and Japan’s NIES use 4.5 SED, and Danish guidance converts to about 4.2.
Type VRarely burns7.5SED50 min at UV 10Thin evidence

Evidence base: 2 study samples, 13 people, one research group.

Why 7.5 SED

What argues against it

  • Thirteen people is a thin base, and the two averages differ twofold. Japan’s NIES uses 6 SED.
Type VIVery rarely burns12SED80 min at UV 10Very thin evidence

Evidence base: 1 study sample, 6 people.

Why 12 SED

  • No study measured 12 SED. It is a deliberate choice below the one measured average, 15.2 SED in six people (Shih 2018), whose individuals ranged from 8.3 to 20.5.
  • Redness is harder to see on dark skin, so measured figures may run high (Fajuyigbe & Young 2016). Two participants in a US FDA study reddened at 6.0 and 11.75 SED (Coelho 2013).

What argues against it

  • Six people is not enough to call any figure typical. Some people redden below 12 SED, and Japan’s NIES uses 10.

Time to the limit is time in full sun at UV Index 10 with no sun yet that day, no shade and no sunscreen. If you skip the skin type question, Sunwize uses type III.

How each limit sits against the evidence

Each row is one skin type. Dots are averages from phototest studies, diamonds are figures from health authorities and standards bodies, and the band is the indicative range from ICNIRP, the international radiation protection commission.

2357.5101520UV dose in SED (log scale)Type I2Type II2.5Type III3.5Type IV5Type V7.5Type VI12
2351020UV dose in SED (log scale)Type I2Type II2.5Type III3.5Type IV5Type V7.5Type VI12
Show the chart’s numbers as a table
TypeMeasured study averages (SED)Official figures (SED)ICNIRPLimit
I2.1 Shih 2018 (6 people); Shih 2020 (12 people)
2.2 Valbuena 2020 (5 people; weighting not stated)
2.3 Pérez Ferriols 2014 (4 people)
2 ARPANSA; Health Canada; DIN 5050 and German sunbed law; Japan NIES
1.7 Danish Cancer Society (converted from minutes)
under 22
II2.4 Valbuena 2020 (41 people; weighting not stated)
2.6 Shih 2018 (6 people); Shih 2020 (14 people)
2.7 Lee 2024 (3 people, back)
2.8 Pérez Ferriols 2014 (89 people); Miller 2012 (back)
3.7 Felton 2016 (10 people)
2.5 DIN 5050 and German sunbed law; Health Canada; Japan NIES; Australian position statement (skin types 2 to 3)
2.3 Danish Cancer Society (converted from minutes)
2 to 32.5
III3.1 Miller 2012 (back); Lee 2024 (11 people, back)
3.2 Shih 2018 (7 people)
3.5 Pérez Ferriols 2014 (108 people)
3.6 Valbuena 2020 (48 people; weighting not stated)
4.2 Shih 2020 (19 people)
3.5 DIN 5050 and German sunbed law; COST-713 (EU)
3 Japan NIES
2.7 Danish Cancer Society (converted from minutes)
2.5 Australian position statement (skin type 3, dose on an upright body)
3 to 53.5
IV4.8 Valbuena 2020 (19 people; weighting not stated)
5.1 Pérez Ferriols 2014 (31 people)
5.6 Shih 2018 (7 people)
5.7 Shih 2020 (6 people)
4.5 DIN 5050 and German sunbed law; Japan NIES
4.2 Danish Cancer Society (converted from minutes)
5 to 75
V7.5 Shih 2018 (7 people)
14.6 Felton 2016 (6 people)
8 German sunbed law
6 Japan NIES
5.8 Danish Cancer Society (types 5 and 6, converted from minutes)
7 to 107.5
VI15.2 Shih 2018 (6 people)10 Japan NIES; German sunbed law (“over 10”)over 1012
Hover over or tap a mark for its sources. ICNIRP gives “under 2” for type I and “over 10” for type VI, drawn to the chart’s edge. Danish figures are converted from published minutes.

Cancer Council Australia publishes no dose per skin type, so we cite it for its advice to protect your skin whenever the UV Index is 3 or more, and not for any limit.

How the dose is counted

UV IndexApple Weather’s hourly forecast for where you are
Time outsideChecked every 30 seconds
Sunscreen and shadeSunscreen you log, shade you report, and nearby buildings and trees where they’re mapped
Your doseIn SED, added up through the day
9 SED an hourat UV Index 10 in full sun (1 UV Index = 0.025 W/m²)
Half the UVwhen you tell Sunwize you’re in the shade, which blocks direct sun but not UV scattered from the sky

A 30-second stretch counts in full when Sunwize judges you more likely outside than in, and not at all otherwise. It never adds part-doses for uncertain time, so a long, unclear day can’t pile up dose that never happened.

Why we count the dose on a flat surface in full sun

The UV Index describes UV falling on a flat, upward-facing surface. That’s what your most exposed skin gets. Most of the rest gets a fraction of it.

100%Top of the head30%Forehead80%Tops of shoulders50%Forearms
Share of the counted dose, standing. Averaged over the whole body with the sun high overhead, skin gets about 15%. Lying down, anything facing up gets about 100%. Sources: Schmalwieser 2025, Downs & Parisi 2012, Schmalwieser 2023.

For the face, arms and legs of someone walking around clothed, the count overstates the skin dose by about two to five times. We don’t lower it, for two reasons. The margin doesn’t hold for everyone: bare shoulders, the scalp, the tops of the feet and anyone lying down get the full count. And reading low is the dangerous error: students given UV meters that under-read stayed out longer and burned more often (Carli 2008).

The day adds up

Your limit is for the whole day. Going inside pauses the count. It doesn’t reset it.

Inside: paused0%25%50%75%100%Your limitHeads-up7 am9 am11 am1 pm3 pm5 pm7 pm80% Sunwize35% if reset
Paused0%25%50%75%100%Your limitHeads-up7 am10 am1 pm4 pm7 pm80%35%
An illustrative day, as a share of the day’s limit. At midnight the count starts again from zero.

Why a break indoors doesn’t reset it

  • Redness follows the total dose. Across 13 human studies from 1927 to 1989, “in every case, reciprocity for erythema was shown to hold” (SCCP 2006).
  • Doses a day apart still add up. Doses too small to redden, given 24 hours apart, lowered the dose needed for redness (Kaidbey & Kligman 1981). A lunch break is far shorter than a day.
  • Redness shows up late. It appears 1 to 6 hours after exposure and peaks about a day later (ICNIRP 2004), so skin that looks fine at lunch says nothing about the morning.
  • Repair is slow. Half of the DNA damage that tracks sunburn is still there about 33 hours later (Young 1996).
Time inside (2–8 h)0%25%50%75%100%0 h8 h24 h48 h72 hHours since the exposureNext morning85%61%37%
Time inside (2–8 h)0%25%50%75%100%0 h8 h24 h48 h72 hHours since the exposureNext morning85%61%37%
Show the chart’s numbers as a table
AfterNot yet repaired
2 h (a short break)96%
4 h (a long lunch)92%
8 h (an afternoon inside)85%
24 h (the next morning)61%
48 h (two days later)37%
72 h (three days later)22%
Share of a morning’s DNA damage (CPDs, the lesion that best tracks sunburn) not yet repaired. Our arithmetic on the half-life measured in fair skin by Young 1996. Repair speed varies up to 20-fold between people (Bykov 1999).

What health authorities say

  • Cancer Council: “the longer you’re exposed, the more the damage adds up” (Cancer Council).
  • US FDA: SPF “is not directly related to time of solar exposure but to amount of solar exposure” (FDA).
  • Exposure standards set their limits as a sum over the day (ICNIRP 2004, EU Directive 2006/25/EC), and ARPANSA places skin’s recovery in the “normal 24 hours cycle of light and dark” (ARPANSA RPS12).

Why the count starts again at midnight

Every exposure standard uses a daily window, and in a 2024 trial of the Shade research dosimeter, “participants’ daily UV exposure would reset to zero at midnight” (Dumont 2024). Midnight is a convention, not a recovery, and that has a cost we list under known limitations.

Sunscreen

Sunwize counts a third of the SPF on the bottle, for 2 hours from when you log it.

SPF 15
On the label, 6.7%Sunwize counts 20%
SPF 30
On the label, 3.3%Sunwize counts 10%
SPF 50
On the label, 2%Sunwize counts 6%
Share of the UV that gets through sunscreen, while it’s counted: 2 hours from when you log it.
  • Why a third. SPF is measured under a thick layer that people rarely put on. On one holiday, SPF 15 as people applied it worked like about SPF 3 (Petersen 2013).
  • Putting more on protects the time ahead. It never gives back dose already counted, because SPF is about the amount of UV, not time (FDA).
  • Why we don’t credit more. Generous credit keeps people out longer: higher-SPF sunscreen lengthened sunbathing in a blinded trial (Autier 1999).

Warnings

A heads-up at 75% of your limit or 15 minutes before it, whichever comes first, and a warning when you reach it.

  • Both are worked out from the day’s total, never from one outing.
  • After 10 minutes inside, the heads-up can come again on your next trip out.
  • The share of people who’d already show faint pink when the heads-up falls due is about the same for every skin type, so it means the same thing for everyone.
Type I
8–14%
Type II
11–20%
Type III
11–20%
Type IV
10–17%
Type V
15–22%
Type VI
2–7%
Share of people of each type who would already show faint pink at 75% of their limit, if all the counted dose reached untanned skin. Modelled from study averages and spreads; no study reports it directly. At Australian summer UV the 15-minute rule usually comes first for fair skin, so these are upper bounds.

Known limitations

What Sunwize can’t do, and where the evidence runs out. Please read these before you rely on a number.

The limit

It’s a typical figure, not yours

People with the same skin type differ two- to fourfold. Each limit sits near the average, so many people show faint pink before 100%: about 4 in 10 with type I, about half with types II to IV, a third to a half with type V and a fifth to a quarter with type VI.

It isn’t a safe amount

DNA damage is measurable from a fifth of the first-redness dose in every skin type, and immune effects from a quarter to a half. The limit is about sunburn. It says nothing about skin ageing or skin cancer risk.

No trial shows alerts prevent sunburn

Two trials of UV alerts set by skin type found no difference in sunburn (Hacker 2018, MELD 2023). Sunwize can’t promise it prevents burns.

Little evidence for brown and dark skin

The limits for types V and VI rest on 13 and 6 volunteers from one research group. Redness is also harder to see on dark skin, so the measured figures may run high.

You choose your skin type

People tend to rate their skin darker than it is (Reeder 2010). Pick one type too dark and you’d show faint pink before your limit about 9 times in 10.

Nothing carries over between days

The count starts from zero at midnight, but about 60% of a day’s DNA damage is still there the next morning. In fair skin, two or three days running at about two-thirds of the limit can bring clear redness. It matters most early in the season and at the start of a beach holiday.

The count and the warnings

The UV is a forecast, not a measurement

Sunwize uses Apple Weather’s hourly UV forecast for where you are. We know of no study that checks it against ground measurements in Australia, and an hourly forecast can’t follow a cloud passing over.

Where you are is an estimate

Sunwize works out from your phone whether you’re outside. It can miss time outside or count time you spent inside, so check the timeline and correct it.

Some skin burns sooner than the count suggests

Bare shoulders, the scalp, the tops of the feet and anyone lying down get the full counted dose. Sweat and salt water lower the threshold by about 15 to 30%, and early in summer skin hasn’t yet built up any tolerance.

Sunscreen is a fixed estimate

Sunwize can’t see how much you put on, the spots you missed or whether it has rubbed or washed off. It counts a third of the label for 2 hours whatever happens.

Warnings and estimates have gaps

Head back out already above 75% and the heads-up can arrive only a few minutes before your limit. The limit warning comes once a day: after it, more sun shows on the gauge but sends no new notification. Some time estimates, such as the forecast’s, assume no sun yet that day.

What the research can’t tell us yet

  • No study has split one sunburning dose by hours indoors. The case for adding up within a day is indirect, though every line of evidence agrees.
  • The share of people who redden before each limit is modelled from study averages and spreads. No study reports it directly.
  • The carry-over figures are arithmetic on one repair rate measured in fair skin, not measurements in real sun.
  • Several key papers were read as abstracts only, including the two largest skin-type samples (461 and 652 people).
  • No study has checked any app’s sunburn warning against sunburns people actually got.

Sources

Measured first-redness doses

Official and published figures

Sunburn, body sites and skin type

The dose over a day and across days

Alerts, sunscreen and behaviour