FDA-Cleared vs
Over-the-Counter
Pulse Oximeters
A cleared pulse oximeter went through 510(k), including a supervised low-oxygen test. A wellness or sports meter never did. The code is not on the box, and the price will not tell you which one you bought.
A fingertip oximeter that went through 510(k) review sat through a supervised low-oxygen test on volunteers, compared with blood samples. FDA files those as DQA or OLK. PGJ is a wellness meter. OCH is a sporting or aviation meter. Those two never got that test. The code is not on the box, and the price will not tell you which one you bought.
They all clip on the same way. They all light up with the same confident two-digit number. Here is what FDA says about the ones sold as general wellness or sporting and aviation products. They "are not intended for medical purposes, so they do not undergo FDA review," and they "are not evaluated by the agency for use in clinical decision-making or determining whether to seek medical intervention."1
Same Rule, Different Review
Medical-use oximeters are Class II devices under 21 CFR 870.2700.2 That covers the unit on a hospital monitor and the small set of over-the-counter units FDA has actually cleared. Sell the same clip as a wellness gadget, or as a sporting and aviation accessory, and it sits under a different code. FDA never reviewed it, and FDA does not want that number used to decide if someone needs care.1 Nobody has to print the code where a buyer will see it.
| Code | What FDA calls it | Premarket review | What to ask for |
|---|---|---|---|
| DQA | Oximeter | 510(k) required | The K number, plus the accuracy figure and the conditions it was measured under |
| OLK | Pulse oximeter, over the counter | 510(k) required | The same file. Over the counter does not mean the device skipped review |
| NLF | Oximeter, reprocessed | 510(k) required | The K number on the reprocessed device, not on the original |
| PGJ | Oximeter, wellness | No 510(k). Enforcement discretion | Do not use the reading for a medical decision |
| OCH | Oximeter, infrared, sporting and aviation | No 510(k). Enforcement discretion | FDA describes these as exercise heart-rate monitors. Do not use the reading for a medical decision |
Look up the code first. DQA and OLK went through 510(k). PGJ and OCH hit the shelf with no accuracy file you can open. After that, read what the maker prints for motion and for poor blood flow. I have watched people try to sort this by price. It doesn't work.
Enforcement discretion means FDA decided not to ask for a submission, so there's no accuracy file and they don't have to produce one if you call. FDA describes the OCH group as devices "solely for use with sporting and aviation activities," meant "to monitor heart rate during exercise."2 Zacurate files its Pro Series 500DL under that heading.3 The label matches the code. I keep seeing the other half of that sale. Dad comes home on two liters, and his daughter orders the nineteen-dollar unit because the photo looked medical. She was not shopping for an aviation accessory.
Sellers give themselves away on the vocabulary. A pulse oximeter gets a 510(k) when FDA agrees it's close enough to one already cleared. Premarket approval is a different path, the one for the highest-risk devices, and it takes clinical trial evidence.4 An honest listing says "510(k) cleared," gives the K number, states an accuracy figure, and says whether that figure was a still hand, a moving hand, or a poorly perfused one. A seller who writes "FDA approved" either doesn't know the difference or is counting on you not to.
What FDA Clearance Actually Proves
Clearance doesn't mean the next patient will read true. A 2026 lab study put 34 fingertip oximeters on healthy volunteers and dropped their oxygen on purpose. Twenty-two of the 34 met the 2013 FDA mark, an ARMS of 3 percent or better. One of the 34 met both parts of the standard FDA has drafted, the accuracy mark and the test for bias between light and dark skin.5 Cleared devices miss. FDA has spent two years trying to raise the bar.
The 2013 standard asks for at least ten volunteers and an average error of 3.0 percent or better across the range that matters. It also asks for "at least 2 darkly pigmented subjects or 15% of your subject pool, whichever is larger."6 Draft guidance from January 2025 would ask for bigger groups, and for skin tone measured on a published scale instead of eyeballed.7 That guidance is still a draft. A cleared device on the shelf today was held to 2013.
Clearance still buys a file. Somebody ran the test, wrote the numbers down, and FDA looked at them. You can look the K number up yourself. A wellness meter doesn't hand you a worse number. It hands you no number, and you can't correct an error nobody measured.
What I want in print is the miss when the patient is moving, and the miss when the finger is poorly perfused. I want the weakest pulse the device claims it can still read, how many people were in the test, and what their skin tones were. One accuracy percentage with no conditions on it is not a spec I can use on a call.
What That Accuracy Percentage Means
When a sheet says 2 percent, that is the average miss across everybody in the test. Some readings landed closer than 2 points. Some landed farther. Your next patient is not promised a reading inside 2 points. That average is called ARMS, accuracy root mean square, and it's useless until the maker says whether the hand was still, moving, or shut down.
What Happens When Nobody Checks
A Dutch group ran ten popular consumer fingertip oximeters against arterial blood gases on 35 ICU patients. None of the ten met a 3 percent average error, which is the mark that study used. Depending on the device, the average error ran from 0.6 points low to 4.8 points low.8
They also scored the meters on low oxygen, a blood level at or below 90 percent. A normal reading ruled that out 98 to 99 percent of the time. A low reading was actually that low only 11 to 30 percent of the time.8 Both figures came off ICU patients with arterial lines, and only about 5 percent of the blood samples were that low. I would not carry either percentage into a living room and pretend it still holds.
What This Looks Like in a House
This is the report I pull out when someone tells me a cheap meter is close enough. In March 2023 a 76-year-old man filed it with FDA. His sporting-and-aviation oximeter, product code OCH, swung from 132 down to 32 while he was checking oxygen and pulse. On another check it showed 113 while two other meters read 62 and 69. He called paramedics. The emergency physician found nothing wrong, and the company told him the meter sounded like it was going bad.9 A saturation does not display 132 or 113. He did not say whether 62 and 69 were oxygen or pulse. You cannot subtract a steady fudge factor from noise. It took a truck and an ED bed to prove the number was junk.
Skin Pigmentation
The miss is not random. A 2020 letter in the New England Journal of Medicine went looking for hidden low oxygen: a blood level under 88 percent while the screen sat at a comfortable 92 to 96. That showed up in 11.7 percent of paired readings from Black patients and 3.6 percent from White patients.10 Across 87,971 encounters at 215 hospitals the same pattern was 6.9 percent and 4.9 percent, and those patients had more organ failure and higher mortality.11 A Veterans Health Administration study drew the line in a looser place and got 19.6 percent and 15.6 percent.12 Move the cutoff and the ratio of Black patients to White patients slides from about 3 down to about 1.3. If you quote a percentage, quote the cutoff that made it.
The device on the finger changes the size of that gap. One lab put three brands on the same volunteers, same day, same room. Light skin versus dark skin came out 2.35 percent on the Nonin, 3.38 percent on the Novametrix, and 4.30 percent on the Nellcor.13 The people did not change. The hardware did.
UCSF did it again in 2024 with eleven fingertip oximeters and 34 volunteers. Four were 510(k) cleared. Nine cost under sixty dollars. Five missed the 3 percent mark. Nine did worse on the darkest-skinned volunteers than on the lightest.14 No manufacturer paid for the study.
Two Meters That Publish the Working Numbers
We stock two that print the motion figure and the poor-perfusion figure.
The Nonin Onyx Vantage 9590 holds 510(k) K112843. Nonin publishes a 2 percent average error at the finger or thumb, and the same 2 percent when blood flow is poor, down to a pulse strength of 0.3 percent. Toes are rated at 3 percent. They say it has been tested while the subject was moving, and they do not publish a motion figure, so I will not tell you the 2 percent covers a shivering patient.15 Finger size and the cleaners that were tested are on that page.
The Masimo MightySat Rx 9941 sits on one cleared platform, 510(k) K181956. The 9709 shows oxygen, pulse rate, and perfusion index. The 9941 adds PVi and a respiration rate on that same oximetry. Masimo publishes 2 percent when the hand is still, 3 percent while it is moving, and 2 percent with poor blood flow. That poor-flow figure came off a bench simulator, not a patient. The respiration rate is cleared for adults only, 4 to 70 breaths a minute, and it is the drive to breathe. It is not proof air is moving. PVi is not in the cleared uses at all.16 If the patient is bouncing in the back of the truck, the motion number is the one I want in print, and this is a fingertip unit that actually prints one.
When you need to know air is leaving the lungs, that reading is exhaled carbon dioxide. The CMI PC-900B reads EtCO2, SpO2, pulse, and a respiration rate on one handheld.
Where Each Reader Gets Caught
The same bad meter embarrasses different people for different reasons. Read the line on the sheet that matches the person holding it.
| Reader | Where an unreviewed unit fails | What to require |
|---|---|---|
| Prehospital | A shut-down pressure in the back of the truck. Poor blood flow is a normal call, and the truck is moving. | A published low-perfusion figure and the weakest pulse it claims it can read, the way the Onyx Vantage 9590 prints 0.3 percent. A sheet that only describes a still, warm hand is describing somebody else. |
| Home health | The same patient, visit after visit. A small steady bias becomes the baseline the whole team believes. | The same model the ordering clinician uses, so the office and the field are trending one instrument. If you cannot match it, write the device name next to every reading. |
| Family caregiver | One number at two in the morning, and it decides whether anyone gets called. | A K number you looked up, on a cleared fingertip pulse oximeter, with a sensor that fits. Write the time down. The clinician needs the trend more than the last digit. |
Three Checks Before You Trust the Number
1. The K Number Is Public
Ask for the 510(k) number and look it up in FDA's premarket database.17 If they come back with "FDA registered," treat it as unreviewed. Registration means the company filled out a form and paid a fee so FDA knows the building exists. Clearance means FDA looked at that device and its test data.4 When you want a number that did not come from the sales sheet, the UCSF Open Oximetry Project posts lab results by device, free.18
2. What a Two-Wavelength Oximeter Cannot See
A standard pulse oximeter uses two wavelengths. It can tell oxygen-carrying blood from oxygen-poor blood. It cannot see carbon monoxide, so a poisoned patient can sit at 99 percent and get worse in front of you. Methemoglobin parks the display in the mid-eighties no matter what the blood is doing.19 On a fire ground, a generator in the garage, or a patient who just got topical numbing spray, that blind spot is the whole assessment. SpCO is the blood reading that shows the carbon monoxide, and a fingertip SpO2 meter does not display it. The kit for that job is the Rad-57 with SpCO, adult and pediatric probe, part 3735. The Rainbow sensors for those sizes are the adult DCI, part 2696 and the pediatric DCI-P, part 2697. A lab panel still confirms it. Nail polish, a poor pulse, motion, a low red-cell count, and a bright room light will also lie to you, and those are the ordinary ones.1
Carbon monoxide is one of the poisoning causes on the altered-patient checklist.
3. The Trend, Then the Patient
FDA wants the trend, not one absolute number.19 A trend is only readable if the error stays put. Cheap units get worse as the patient drops, which is the range where you start caring about the direction.8
If the device shows a waveform or a perfusion index, look at it before you chart the saturation. A ragged trace means I am not writing that number down. When the number looks fine and the patient doesn't, the patient wins. Work of breathing, mental status, the skin, and whether they can finish a sentence were how you judged the patient long before any of us clipped a meter on a finger.
Frequently Asked Questions
Is an over-the-counter pulse oximeter the same thing as an uncleared one?
No. Some over-the-counter oximeters are FDA cleared under product code OLK. The ones that were never reviewed are filed as wellness (PGJ) or sporting and aviation (OCH).2 The filing says what the device is for, and that is what splits them. A prescription at the counter is a separate question.
How can a buyer confirm a fingertip pulse oximeter is FDA cleared?
Ask the seller for the 510(k) number. It starts with K and six digits follow. Look it up in FDA's premarket database.17 A cleared device has a file there with the maker, the device name, and the date. No K number means nobody reviewed the accuracy, whatever the listing claims.
Is "FDA registered" the same as FDA cleared?
No. Registration is a form and a yearly fee so FDA knows the building exists. Nothing about the device gets tested. Clearance means FDA reviewed that device and its data.4 A K number that matches the database is the one I trust.
Are more expensive fingertip pulse oximeters more accurate?
Price is a weak tell. Twelve of the 34 fingertip oximeters in that 2026 desaturation study missed the 2013 FDA accuracy mark.5 I look for a motion figure, a poor-perfusion figure, and a description of who was in the test. A higher price with none of those lines doesn't grow them.
What does ARMS mean on a pulse oximeter specification sheet?
ARMS stands for accuracy root mean square. It's the averaged gap between the screen and the blood sample, across the readings in the test. FDA's 2013 guidance asks for 3.0 percent or better.6 That number describes the pile of readings. It doesn't promise your next reading lands inside it.
Does nail polish affect pulse oximeter readings?
FDA lists nail polish with the other things that throw a reading off: poor circulation, skin temperature, skin thickness, and skin pigment.1 Dark polish soaks up the light the sensor is trying to read. Take it off that finger, or move to a bare one.
Why does a pulse oximeter read low or fail on cold hands?
The meter is reading the pulse in the finger. A cold finger, or a low-pressure one, may not have a pulse left to read. Warm the hand. Switch fingers. If you need the number anyway, use a device that publishes a poor-perfusion figure and states the weakest pulse it can still see. Nonin rates the Onyx Vantage 9590 down to 0.3 percent pulse strength.15
Can a pulse oximeter rule out carbon monoxide poisoning?
No. A standard two-wavelength meter cannot tell carbon monoxide from oxygen. A bad exposure can sit there with a normal or high reading while the tissues are starving.19 SpCO is a separate reading. The handheld that displays it, with an adult and a pediatric probe, is the Rad-57, part 3735, used with the Rainbow adult DCI, part 2696 or the pediatric DCI-P, part 2697. On a fire or a generator call, the scene still outweighs the screen, and a lab confirms it.
What SpO2 qualifies a patient for home oxygen under Medicare?
Group I is an arterial PO2 at or below 55 mm Hg, or a saturation at or below 88 percent, at rest, during sleep, or during exercise, on the terms in the coverage rule. Group II is a PO2 of 56 to 59 mm Hg, or a saturation of 89 percent, and only when the patient also has dependent edema suggesting heart failure, pulmonary hypertension or cor pulmonale, or a hematocrit over 56 percent. The treating practitioner has to order the test and review it. An equipment supplier does not count as the lab.20
Can one fingertip pulse oximeter be shared between patients?
On a truck or in a clinic, yes, if the maker published a cleaning method the device can survive. The manual lists the cleaners they actually tested.15 A consumer unit with no tested cleaning instructions doesn't get passed from patient to patient. That is infection control. It has nothing to do with whether the saturation is right.
Both of these print their accuracy under stated conditions, and both have a K number you can look up. K112843 is the Nonin Onyx Vantage 9590. K181956 is the Masimo MightySat Rx 9941.
Carry a meter that was tested
K112843 is the Nonin Onyx Vantage 9590. K181956 is the Masimo MightySat Rx 9941. For carbon monoxide on the blood, the SpCO kit with an adult and a pediatric probe is the Rad-57, part 3735, with the Rainbow adult DCI, part 2696 and the pediatric DCI-P, part 2697.
References
- US Food and Drug Administration. Pulse oximeter basics. FDA Consumer Updates. Content current as of March 26, 2025. Accessed October 17, 2026. https://www.fda.gov/consumers/consumer-updates/pulse-oximeter-basics
- US Food and Drug Administration. Product classification database: devices under 21 CFR 870.2700 (product codes DQA, NLF, OLK, PGJ, OCH). Accessed August 11, 2026. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpcd/classification.cfm?start_search=1®ulationnumber=870.2700
- Zacurate. Pro Series 500DL fingertip pulse oximeter (listed under athletic and aviation pulse oximeters). Accessed August 11, 2026. https://zacurate.com/products/zacurate-pro-series-500dl-fingertip-pulse-oximeter
- US Food and Drug Administration. Learn if a medical device has been cleared by FDA for marketing. Content current as of December 29, 2017. Accessed August 11, 2026. https://www.fda.gov/medical-devices/consumers-medical-devices/learn-if-medical-device-has-been-cleared-fda-marketing
- Hughes C, Chen D, Law T, Bickler P, Feiner J, Shmuylovich L, et al. Pulse oximeter performance and skin pigment: comparison of 34 oximeters using current and emerging regulatory frameworks. Anesth Analg. Published online April 20, 2026. doi:10.1213/ANE.0000000000008048
- US Food and Drug Administration. Pulse oximeters: premarket notification submissions [510(k)s]: guidance for industry and Food and Drug Administration staff. Issued March 4, 2013. Accessed August 11, 2026. https://www.fda.gov/media/72470/download
- US Food and Drug Administration. Pulse oximeters for medical purposes: non-clinical and clinical performance testing, labeling, and premarket submission recommendations. Draft guidance for industry and FDA staff. January 2025. Docket FDA-2023-N-4976. Accessed August 11, 2026. https://www.fda.gov/media/184896/download
- Harskamp RE, Bekker L, Himmelreich JCL, et al. Performance of popular pulse oximeters compared with simultaneous arterial oxygen saturation or clinical-grade pulse oximetry: a cross-sectional validation study in intensive care patients. BMJ Open Respir Res. 2021;8(1):e000939. doi:10.1136/bmjresp-2021-000939
- US Food and Drug Administration. MAUDE adverse event report: Innovo Medical fingertip oximeter model IP900AP. MDR report key 16645764. Event date March 27, 2023. Product code OCH. Accessed October 17, 2026. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfmaude/detail.cfm?mdrfoi__id=16645764&pc=OCH
- Sjoding MW, Dickson RP, Iwashyna TJ, Gay SE, Valley TS. Racial bias in pulse oximetry measurement. N Engl J Med. 2020;383(25):2477-2478. doi:10.1056/NEJMc2029240
- Wong AI, Charpignon M, Kim H, et al. Analysis of discrepancies between pulse oximetry and arterial oxygen saturation measurements by race and ethnicity and association with organ dysfunction and mortality. JAMA Netw Open. 2021;4(11):e2131674. doi:10.1001/jamanetworkopen.2021.31674
- Valbuena VSM, Seelye S, Sjoding MW, et al. Racial bias and reproducibility in pulse oximetry among medical and surgical inpatients in general care in the Veterans Health Administration 2013-19: multicenter, retrospective cohort study. BMJ. 2022;378:e069775. doi:10.1136/bmj-2021-069775
- Bickler PE, Feiner JR, Severinghaus JW. Effects of skin pigmentation on pulse oximeter accuracy at low saturation. Anesthesiology. 2005;102(4):715-719. doi:10.1097/00000542-200504000-00004
- Leeb G, Auchus I, Law T, Bickler P, Feiner J, Hashi S, et al. The performance of 11 fingertip pulse oximeters during hypoxemia in healthy human participants with varied, quantified skin pigment. EBioMedicine. 2024;102:105051. doi:10.1016/j.ebiom.2024.105051
- Nonin Medical. Onyx Vantage model 9590 finger pulse oximeter operator's manual. Part no. 113634-001-01. Accessed August 11, 2026. https://www.nonin.com/wp-content/uploads/113634-001-01_ENG.pdf
- Masimo Corporation. MightySat Rx fingertip pulse oximeter operator's manual. LAB-12373B. Accessed August 11, 2026. https://techdocs.masimo.com/globalassets/techdocs/pdf/lab-12373b_eifu.pdf
- US Food and Drug Administration. 510(k) premarket notification database. Accessed August 11, 2026. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm
- Open Oximetry Project, University of California San Francisco Hypoxia Research Laboratory. Pulse oximeter performance database. Accessed August 11, 2026. https://openoximetry.org/oximeters/
- US Food and Drug Administration. Pulse oximeter accuracy and limitations: FDA safety communication. Issued February 19, 2021. Updated November 16, 2023. Accessed August 11, 2026. Archived at https://public4.pagefreezer.com/content/FDA/20-02-2024T15:13/https://www.fda.gov/medical-devices/safety-communications/pulse-oximeter-accuracy-and-limitations-fda-safety-communication
- Centers for Medicare and Medicaid Services. National coverage determination: home use of oxygen (240.2). Effective September 27, 2021. Accessed October 17, 2026. https://www.cms.gov/medicare-coverage-database/view/ncd.aspx?NCDId=169
This article is for trained clinicians and for caregivers acting under a clinician's direction. It does not replace local protocols, medical direction, formal clinical training, or the instructions of the prescribing clinician. Clinicians should practice within their scope and their service's current guidelines.