Confined Spaces: Is 19.5 Percent Oxygen Really Safe?

June 1, 1999
Everybody knows that an oxygen level of 19.5 percent is safe for entry into confined spaces. Well, once again, what everybody knows is simply wrong!

I frequently ask participants in my confined space courses, "What makes 19.5 percent so special? Why isn't the acceptable level of oxygen something else, like 19.3 percent, 19.8 percent or 20.2 percent?" The usual answer I get is, "Because OSHA says so."

Contrary to popular belief, OSHA doesn't say that an oxygen level of 19.5 percent is "safe." Instead, 29 CFR 1910.146(b) defines a hazardous atmosphere as one "... that may expose employees to the risk of death, incapacitation, impairment, or ability to self-rescue (that is, escape unaided from a permit space), injury or acute illness from one or more of the following causes ...." The standard then goes on to list five causes, one of which is "... atmospheric oxygen concentration below 19.5 percent...."

Note the subtle difference here. The standard does not say that atmospheres containing 19.5 percent are safe; it says that those which have levels below 19.5 percent may be hazardous. While this might sound like hairsplitting, I"ll explain later why it isn't.

Understanding Standards

OSHA standards simply state regulatory requirements. They don't tell us how those requirements should be met, nor do they offer any advice, guidance, or commentary on how to achieve compliance. This is not something unique to OSHA. All codes, regulations and standards, whether issued by NFPA, ANSI or any other standards-making body, presume that readers have a substantial base of technical knowledge related to the subject matter.

Codes, regulations and standards are not intended to be cookbooks that tell us how to do something; but, rather, they are documents that summarize specific expectations. For example, building codes summarize the requirements that structures are expected to meet, but they don"t tell construction workers how to use their tools and equipment. Instead, they essentially presume that craftspeople already have the skills necessary to construct a building and that they will follow the codes to ensure that specific construction goals are achieved.

Similarly, OSHA's confined space regulation presumes that readers have substantial technical knowledge in areas such as, but not limited to, toxicology, fall protection, chemical protective clothing, machine guarding, fire protection, industrial hygiene instrumentation, electrical safety, lockout/tagout, respiratory protection, ventilation and adult learning methods.

In this light, there is a presumption that readers also understand the technical basis for many of the standard's requirements, including the 19.5 percent oxygen value. My experience, drawn from thousands of people who have attended dozens of my courses, suggests that most folks don't have a clue as to why 19.5 percent is significant. While this oxygen level may be acceptable in some situations, relying on it without understanding its basis can lead to fatal consequences.

For example, in the first case I worked on as an expert witness, the entry supervisor for a contract tank cleaning firm tested the atmosphere in a space and found it contained 20.1 percent oxygen. When I saw that number, I was concerned, very concerned. The supervisor wasn't; in fact, he was oblivious to the warning it provided. Instead, he testified with great confidence that he knew that 19.5 percent was "safe." After all, that's what he'd been taught at the confined space course he took at the state fire school.

The next day, three people including the plant safety director died when they entered an oxygen-deficient atmosphere. This, by the way, was the first of three cases I've worked on where one of the people who was killed was the on-site safety officer, but that's a story for another time.

While there were a number of other issues that had a bearing on this tragedy, the fact remains that the entry supervisor had an obvious indication the day before the incident that there was an atmospheric hazard present. An oxygen level of 20.1 percent provided a clear, unambiguous warning that something was wrong. Regrettably, the entry supervisor did not comprehend the warning because he, like so many other people, knew that 19.5 percent was "safe."

To understand why 19.5 percent oxygen may not be an acceptable level for entry into some confined spaces, we need to know something about the respiratory system.

The Respiratory System

The respiratory system consists of a single airway that branches into smaller and smaller passages, similar to the roots of a tree. At the end are small, grape-like clusters called alveoli. The alveoli are separated from blood-carrying capillaries by cell walls that are permeable to gases, such as oxygen and carbon dioxide. The driving force for the gas exchange across this barrier is a pressure difference that exists on opposite sides of the cell walls. Higher oxygen pressure on one side of the walls allows oxygen to flow from the lungs into the blood, while higher carbon dioxide pressure on the other side of the walls allows it to flow from the blood to the lungs.

Normal atmospheric air at sea level has a pressure of 760 millimeters of mercury (mm Hg). Because air contains approximately 21 percent oxygen, oxygen's contribution to the total pressure, in other words its partial pressure, is 21 percent of 760 mm Hg, or about 159 mm Hg. But as fresh air enters the upper respiratory tract, it is humidified and the water vapor lowers the partial pressure of oxygen to about 150 mm Hg.

Once in the alveolar spaces, oxygen's partial pressure is further reduced by carbon dioxide that has passed from the blood stream to the lungs. Because the pressure of carbon dioxide in the alveoli is about 40 mm Hg, the oxygen's partial pressure drops from 150 to 110 mm Hg.

Once oxygen gets into the blood, it attaches to hemoglobin molecules which carry it to the cells. At an alveolar partial pressure of 110 mm Hg, the hemoglobin molecules are saturated. In other words, they are carrying all the oxygen they can. However, the saturation level is affected by the alveolar partial pressure and a drop in oxygen partial pressure produces a corresponding drop in hemoglobin saturation. It is important to note that physiologists generally agree that the effects of oxygen deficiency begin to manifest at partial pressures of about 60 mm Hg.

Relevance to Confined Spaces

"So what's all this got to do with confined spaces?" you might ask. The oxygen partial pressure inside a confined space may be lower than the 159 mm Hg found in ambient air. If so, the partial pressure of oxygen in the alveolar spaces also will be lower.

If inert gases like argon and nitrogen enter a space, they displace some of the atmospheric air. When this happens, the amount of oxygen and, thus, its partial pressure goes down. For example, assume that nitrogen leaks into a space, lowering the oxygen level to 19.5 percent. The oxygen partial pressure is now 19.5 percent of 760 mm Hg, or 148 mm Hg. When we subtract the partial pressure contributions of water vapor and carbon dioxide, the oxygen partial pressure in the alveolar spaces is down to about 100 mm Hg.

Because the hemoglobin saturation point is 110 mm Hg, the blood is not quite carrying the optimum quantity of oxygen. A partial pressure of 100 mm Hg is still 40 mm greater than the 60 mm physiological danger point. While our margin of safety may be reduced, the situation is not critical.

The 19.5 percent oxygen level that everyone is familiar with is intended to address situations such as this where atmospheric air has been displaced by an inert gas such as argon or nitrogen. However, in light of this, it should be abundantly clear that it's not the percentage of oxygen that's important, but, rather, the partial pressure of oxygen, and that 19.5 percent translates to a partial pressure of 148 mm Hg. Remember, though, all of this is only true at sea level.

Air at high altitudes contains the same percentage of oxygen and nitrogen as air at sea level; however, the barometric pressure at those altitudes is less than that at sea level. For example, the barometric pressure at 5,000 feet is 632 mm Hg vs. 760 mm Hg at sea level. That means that the oxygen partial pressure at 5,000 feet is about 133 mm Hg vs. 160 mm Hg at sea level (21 percent of 632 mm Hg is 133 mm Hg). If we again subtract the contribution for water vapor and carbon dioxide, we will find that alveolar oxygen partial pressure is about 83 mm Hg vs. 110 mm Hg at sea level.

However, at an oxygen level of 19.5 percent the level widely touted as "safe for entry" the oxygen partial pressure in the alveoli drops to about 74 mm Hg. Because the effects of oxygen deficiency will generally manifest at 60 mm Hg, it is clear that the margin of safety under these conditions has narrowed considerably.

While this discussion may appear academic, the effects of decreased oxygen partial pressure becomes an important consideration in some jobs. For example, consider a coastal area tank-cleaning crew that lands a contract to clean tanks in high plains areas such as Denver, Salt Lake City or Albuquerque. When a supervisor tests a space, finds a concentration of 19.5 percent oxygen and says that the space is "safe to enter," is it?

The work crew, unlike the residents of these areas, is not acclimatized, or used to the "thinner" air. After only mild exertion, they may suffer a variety of adverse effects, including reduced peripheral vision, abnormal fatigue and shortness of breath. While these impairments may be inconsequential in ordinary environments, they could impede escape or contribute to fatalities in confined spaces. Do you really think 19.5 percent oxygen is "safe" in this case?

Other Air Contaminants

Another thing I have observed is that most people don't seem to understand that a 1.5 percent drop in oxygen means that a whopping 7.5 percent of something else has gotten into the space.

Recall that, in round numbers, air consists of about 79 percent nitrogen and other gases and about 21 percent oxygen, so the approximate ratio of nitrogen to oxygen is about 4 to 1. This means that, as atmospheric air is displaced from a space, every 1 percent change in the oxygen level will be accompanied by a 4 percent change in the nitrogen level because both gases are displaced at the same rate. In other words, if we start dumping argon into a tank, it won't push out just the oxygen, it pushes out both oxygen and nitrogen in the same proportions that they exist in ambient air, about 4-to-1.

Using round numbers, if the oxygen level drops 1.5 percent from 21 percent to the "safe" level of 19.5 percent, the nitrogen level also must have changed by 6 percent, because four times 1.5 percent is 6 percent. Thus, a total 7.5 percent, or 75,000 parts per million (ppm), of some other substance must be present to cause the oxygen level to drop by just 1.5 percent. If that something else is an inert gas, such as argon or nitrogen, our concern focuses on the partial pressure effects previously explained. But what if it is some other gas or vapor?

Threshold Limit Values for many gases and vapors vary from about 10 to 100 ppm. My favorite solvent ethyl alcohol has the highest TLV, 1,000 ppm, so a level of 75,000 ppm would be 75 times greater than the highest TLV that exists! For substances with TLVs ranging between 10 and 100 ppm, we're now talking between 750 and 7,500 times over the TLV.

While this hazard may be identified through other sampling methods, such as the use of detector tubes, my experience suggests that many people don't understand the magnitude of the problem becuse they don't understand the limitations of the instruments they are using. For example, some participants in my classes tell me they use their combustible gas meters to evaluate the concentration of "toxic" air contaminants, such as acetone, hexane, toluene and methyl ethyl ketone.

Admittedly, these and many other gases and vapors are flammable and can be detected by a combustible gas meter if the concentrations are high enough; however, most combustible gas meters have a detection limit of about 1 percent LEL. This means that, even though the concentration of some gases and vapors may be 10 times over the TLV, the combustible gas meter reads zero. This is because this concentration, as high as it is, still is below the combustible gas meter's limit of detection.

So What's Acceptable?

Many reference sources suggest that air contains 20.95 percent oxygen. However, this value is based on the assumption that the air is "bone dry": in other words, it contains no moisture. However, air in most parts of the country does contain a certain amount of water vapor, which we recognize as humidity. While the exact volume of water that air can hold varies with temperature, a relative humidity of 40 to 60 percent at ambient temperatures can lower the oxygen level by about 0.1 percent. As a practical matter, a value of about 20.8 percent oxygen may be more appropriate than 20.9 percent, because the lower value takes humidity into account.

Now, think about this. If ordinary outside air contains 20.8 percent oxygen, and you're ventilating a space with this air, doesn't it stand to reason that the air in the space ought to also be 20.8 percent? If you make an oxygen measurement and your instrument reads 20.0 percent instead, don't you think you ought to be a little concerned? Shouldnt you ask yourself "Why?"? If you don't know why, should you really let people enter the space?


Contrary to popular belief, 19.5 percent oxygen is not some magic number. Rather, it's a value established on the basis of adverse physiological effects that may manifest at an oxygen partial pressure less than 148 mm Hg. Even if the oxygen is well above 19.5 percent, hazardous concentrations of other gases and vapors may be present. Some gases and vapors may be present at concentrations well above the TLV while, at the same time, they are below a combustible gas meter's limit of detection.

Because ambient air contains about 20.8 percent oxygen, if the oxygen concentration in a space is anything other than 20.8 percent, you should ask yourself "Why?". If you can't come up with a credible answer, you had better not let people enter a space until you can do so.

John Rekus is an independent safety consultant and author of the National Safety Council's Complete Confined Spaces Handbook. With more than 20 years of OSHA regulatory experience, he specializes in conducting OSHA compliance surveys and providing safety seminars for workers and managers. He resides near Baltimore and may be reached at (410)583-7954 or via his web site at

Sponsored Recommendations

Free Webinar: ISO 45001 – A Commitment to Occupational Health, Safety & Personal Wellness

May 30, 2024
Secure a safer and more productive workplace using proven Management Systems ISO 45001 and ISO 45003.

ISO 45003 – Psychological Health and Safety at Work

May 30, 2024
ISO 45003 offers a comprehensive framework to expand your existing occupational health and safety program, helping you mitigate psychosocial risks and promote overall employee...

DH Pace, national door and dock provider, reduces TRIR and claims with EHS solution

May 29, 2024
Find out how DH Pace moved from paper/email/excel to an EHS platform, changing their culture. They reduced TRIR from 4.8 to 1.46 and improved their ability to bid on and win contracts...

Case Study: Improve TRIR from 4+ to 1 with EHS Solution and Safety Training

May 29, 2024
Safety training and EHS solutions improve TRIR for Complete Mechanical Services, leading to increased business. Moving incidents, training, and other EHS procedures into the digital...

Voice your opinion!

To join the conversation, and become an exclusive member of EHS Today, create an account today!