What UV Dose Do You Need for Water Disinfection? A Practical Guide to mJ/cm²
What UV Dose Do You Need for Water Disinfection? A Practical Guide to mJ/cm²
What UV Dose Do You Need for Water Disinfection? A Practical Guide to mJ/cm²
By yoyo-uv — drafted with AI assistance and edited by people
If you specify UV equipment, you will keep seeing the same unit: millijoules per square centimeter, or mJ/cm². People often treat it like one fixed target. It is not. The dose you need depends on the organism, the rule you have to meet, and whether the reactor still delivers that energy in real water at real flow.
This article explains what the unit means, why 40 mJ/cm² and 186 mJ/cm² both show up in official guidance, and what actually changes the dose you get in service.
What UV dose actually is
UV dose is not the electrical wattage printed on a lamp. It is the UV energy that reaches a unit of area. The UK Drinking Water Inspectorate describes it as the product of the intensity from the lamp and the time the water stays in the reactor. The same quantity can be written as mW·s/cm² or as mJ/cm². Those two units are equal.
A bright lamp and a short residence time can give the same dose as a weaker lamp and a longer residence time. If flow rises, time falls, and the dose falls with it unless intensity is raised to compensate.
The number is useful only if you also know the wavelength basis and the water conditions. Official public-system tables in the United States are referenced to 254 nm light from a low-pressure mercury lamp. A different source, or a different water quality, is not automatically the same thing.
Why one number cannot cover every organism
Microbes do not all fail at the same dose. Protozoa such as Cryptosporidium resist ordinary chlorine, but they are quite sensitive to germicidal UV. That is why official protozoa doses sit relatively low once infectivity, rather than a simple in-vitro test, is used as the measure.
Viruses are a different story. Some viruses need much more UV energy for the same log credit. If you size a reactor only around bacteria or Cryptosporidium, you can still miss a virus target.
Where the familiar 40 mJ/cm² figure comes from
For many private supplies and small systems, guidance clusters around 40 mJ/cm². The UK Inspectorate says the minimum dose depends on the organism, but is generally taken to be around 40 mJ/cm².
In North America, that same 40 mJ/cm² figure is the usual Class A design point under NSF/ANSI Standard 55, at 254 nm. British Columbia’s small-system guidance, which quotes the provincial drinking-water officers’ guide, says a Class A system should deliver a minimum UV dose of 40 mJ/cm², hold flow so contact time is kept, and include a sensor, an alarm, and a fail-safe that acts if the dose drops below 40 mJ/cm².
Class B is not a slightly weaker Class A. Class B units are meant as supplemental treatment of water that already meets drinking-water standards. They are not the class used when the source itself may contain pathogens. Mixing the two labels is a common specification error.
The U.S. EPA table is a different kind of number
For public water systems that want inactivation credit under the Long Term 2 Enhanced Surface Water Treatment Rule, the U.S. EPA published a pathogen-specific dose table. The values are referenced to 254 nm from a low-pressure mercury vapor lamp. They apply only to post-filter UV in filtered systems, and to unfiltered systems. The table below is taken from the EPA Ultraviolet (UV) Treatment Toolkit (Table 4, citing 40 CFR 141.720(d)(1)).
Log creditCryptosporidium (mJ/cm²)Giardia lamblia (mJ/cm²)Virus (mJ/cm²)
| 0.5 | 1.6 | 1.5 | 39 |
| 1.0 | 2.5 | 2.1 | 58 |
| 1.5 | 3.9 | 3.0 | 79 |
| 2.0 | 5.8 | 5.2 | 100 |
| 2.5 | 8.5 | 7.7 | 121 |
| 3.0 | 12 | 11 | 143 |
| 3.5 | 15 | 15 | 163 |
| 4.0 | 22 | 22 | 186 |
So 4-log credit for Cryptosporidium or Giardia is 22 mJ/cm². 4-log credit for virus is 186 mJ/cm².
This is the gap that causes most confusion. A 40 mJ/cm² Class A design point sits above the EPA 4-log protozoa values. It sits far below the EPA 4-log virus value of 186 mJ/cm². You cannot read “40 mJ/cm²” as 4-log virus credit under that table.
EPA also requires the reactor itself to be validated, not just labeled. Validation has to account for flow rate, sensor intensity, lamp status, water absorbance, fouling and aging, sensor uncertainty, and the way water actually moves through the chamber. A nameplate dose is not the same as a validated dose.
What changes the dose you actually deliver
Even a well-chosen target fails if the reactor does not deliver it in service.
- Flow rate. If you exceed the maximum design flow, residence time drops. The Inspectorate is direct about this: the maximum design flow rate should not be exceeded.
- Water clarity. Particles can shield microbes from the light. If Cryptosporidium is a concern, UK guidance calls for pre-filtration that can hold turbidity below 1 NTU. Small-unit suppliers often also set a minimum UV transmittance. That limit belongs to the reactor design; it is not one universal number.
- Sleeve fouling. Scale on the quartz sleeve cuts the light that reaches the water. The sleeve has to be cleaned or replaced on a real schedule, not only when the lamp looks dim.
- Lamp age. Output falls over time. UK guidance notes a typical lamp life of about 10 to 12 months, after which output is about 70% of a new lamp. The lamp can still look lit. Light in the window is not proof of dose.
- Warm-up and power loss. If water can pass during warm-up or a lamp or power failure, that water is not disinfected unless a shut-off or an equivalent control stops the flow.
A simple intensity monitor and a fail-safe are strongly recommended on private supplies, even when they are sold as extras on small point-of-use units.
A note on UV-C LEDs and the 254 nm tables
Many compact water devices now use UV-C LEDs rather than a mercury lamp. The EPA dose table is not written for that source. It is referenced to 254 nm light from a low-pressure mercury lamp. Some small-system guidance now discusses UV-LED equipment as a separate design path that still needs validation, or a relevant certification, of the finished system.
If you are integrating compact UVC LED modules into water-treatment equipment, treat wavelength, optical layout, flow, and UV transmittance as a set. Do not copy a 254 nm mercury table onto an LED reactor and assume the credit is the same.
UV also leaves no lasting disinfectant in the pipe. The Inspectorate calls this the principal disadvantage. The water is treated as it passes the light. It is not protected later in a tank or a long main. Hygiene in storage and distribution still matters.
Questions engineers often ask
What UV dose is needed to disinfect drinking water?
There is no single dose for every job. Many private-supply and small-system guides use about 40 mJ/cm² as a practical Class A target at 254 nm. For U.S. public-system credit, EPA lists 22 mJ/cm² for 4-log Cryptosporidium or Giardia, and 186 mJ/cm² for 4-log virus, under the limits in that table.
Is 40 mJ/cm² enough for Cryptosporidium?
That value is higher than EPA’s 4-log Cryptosporidium dose of 22 mJ/cm². The reactor still has to deliver the dose in your water, at your flow, with fouling and lamp age taken into account.
Why is the virus number so much higher?
Some viruses need much more UV energy than protozoa. In the EPA table, 4-log virus credit is 186 mJ/cm².
If the lamp is still on, is the dose still valid?
Not necessarily. Output declines as the lamp ages, and scale on the sleeve blocks light. A lamp can look normal after its useful output has dropped.
What happens if I run above the rated flow?
Contact time shortens, and the delivered dose can fall below the design value. Stay within the validated or rated flow.
About the writer
This article was prepared by YOYO UV, an independently operated company founded in August 2012 in Shenzhen, China, with a team of 31–40 people. We manufacture UV lights and compact modules used in water-treatment equipment. For technical questions, contact Mr. Kevin Pan at info@yoyo-uv.com.