Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
A safety integrity level is a measure of how reliably a protective function must perform on demand, expressed as a band of failure probability under IEC 61508 and IEC 61511.

What Is a Safety Integrity Level?
A safety integrity level, universally shortened to SIL, answers one question: how reliable does this protective function have to be? It is a band, not a score, defined in IEC 61508 and applied to the process industries by IEC 61511, and it attaches to a specific safety function rather than to a device, a system, or a plant.
The distinction matters because vendors sell equipment described as SIL 3 certified. What that actually means is that the device is capable of being used in a SIL 3 loop. The level belongs to the whole function — sensor, logic solver, and final element together, including their diagnostics, proof test interval, and common cause factors.
| SIL | Average probability of failure on demand | Risk reduction factor | Typical application |
|---|---|---|---|
| 1 | 1 in 10 to 1 in 100 | 10 to 100 | Minor equipment damage or limited injury |
| 2 | 1 in 100 to 1 in 1,000 | 100 to 1,000 | Serious injury, significant loss |
| 3 | 1 in 1,000 to 1 in 10,000 | 1,000 to 10,000 | Multiple fatalities on site |
| 4 | 1 in 10,000 to 1 in 100,000 | 10,000 to 100,000 | Catastrophic community impact; very rare in process plant |
These figures describe low demand mode, where the protective function is called on less than once a year — the normal case in process industries. For high demand or continuous mode, typical of machinery and transport, the measure switches to dangerous failures per hour, which is why a railway or automotive SIL conversation sounds different from a refinery one.
The level is an output of hazard analysis, not an engineering preference. The usual sequence runs: HAZOP identifies the hazardous scenario; the consequence and the unmitigated frequency are estimated; existing independent protection layers are credited; and the gap between the residual risk and the tolerable risk determines the required risk reduction. Layer-of-protection analysis is the most common method, with risk graphs used where a quicker judgement is acceptable.
Achieving the level then requires more than a calculation. IEC 61508 distinguishes random hardware failures, which are quantified, from systematic failures, which are controlled by process rigour — specification quality, design review, configuration management, and competence of the people involved. A function can fail its SIL target because the architecture lacks hardware fault tolerance, or because the development process was not disciplined enough, and the second is harder to retrofit.
Once in service, the level is maintained by proof testing at the interval assumed in the calculation. Extending that interval silently, which happens whenever an outage is postponed, degrades the achieved integrity whether or not anyone recalculates it.
This is the point most worth internalising. A SIL figure is a probability derived from failure rates, and it assumes failures are random and independent. An attacker is neither. One compromised engineering workstation can alter the sensor logic, the trip setpoint, and the final element command simultaneously, which is exactly the common cause that the arithmetic excludes.
IEC 62443 handles the adversary with a separate and deliberately non-probabilistic measure, the security level, defined by the capability of the attacker a zone must withstand. The two scales are complementary and must not be conflated: a SIL 3 safety instrumented function in a zone with no access control is not a SIL 3 protection against anything deliberate. Practical guidance, including the HSE’s Operational Guidance OG 86, expects operators to assess both and to show how the security of the zone preserves the assumptions behind the safety calculation.
SIL determination forms part of the demonstration of reduction to as low as reasonably practicable that underpins a COMAH safety report, and the Health and Safety Executive will look at how levels were derived, how proof testing is evidenced, and whether cyber threats to the protective layers have been considered. Where the same site is an operator of essential services, the NCSC’s Cyber Assessment Framework runs alongside, and the two assessments increasingly ask about the same equipment from opposite directions.
At Privia Security, our Cyber Risk Analysis Services connect security assessment to the process consequences your safety studies have already quantified.
Rarely in the process industries. Achieving it demands extreme hardware fault tolerance and development rigour, and most operators find it cheaper and more reliable to reduce the underlying hazard or add an independent non-instrumented layer such as a relief device. SIL 4 appears more often in rail and nuclear contexts where the consequence model differs.
That the device has been assessed against IEC 61508 and has published failure rate data, diagnostic coverage, and systematic capability allowing it to be used in a loop up to a stated level. It does not mean a loop built from it achieves that level: architecture, proof test interval, and common cause factors all affect the result.
Directly. The probability of failure on demand is calculated assuming a specific test interval, so a function proof tested every two years performs differently from the same function tested every five. Deferring tests without recalculating quietly moves the achieved integrity below the target, and auditors increasingly ask for evidence of the actual interval.
Yes, because the calculation assumes independent random failures and an attacker is a deliberate common cause. If one engineering workstation can write to the logic solver, the redundancy in the loop offers no protection against someone who changes the application. That is why security levels are assessed separately and why safety zones are segregated.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.