Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
CRL and OCSP are the two mechanisms for telling a client that a certificate was revoked before its expiry date. Both work in theory; in practice revocation checking has mostly been abandoned.

What Are CRL and OCSP?
Certificates carry an expiry date, but sometimes an authority needs to withdraw one earlier: the private key leaked, the domain changed hands, or the certificate should never have been issued. Revocation is how that message reaches clients, and it has two classic forms. A Certificate Revocation List is a signed file listing the serial numbers an authority has revoked. The Online Certificate Status Protocol lets a client ask about one certificate and receive a signed answer of good, revoked, or unknown.
Both are specified, both are deployed, and neither has ever worked well.
RFC 5280 defines the CRL format, and the certificate points at one through its CRL Distribution Point extension. The client downloads the list, checks the serial number, and caches the result until the next scheduled update.
The arithmetic breaks down at scale. A large authority may have millions of unexpired certificates, and a mass revocation event turns the list into a very large download that every client fetches at once. After Heartbleed was disclosed on 7 April 2014, CloudFlare reissued and revoked certificates in bulk, and the resulting CRL growth cost an estimated hundreds of thousands of pounds in bandwidth across the internet. A list that is downloaded daily also means a revocation can take a day to take effect.
OCSP, defined in RFC 6960, replaced the bulk download with a single question. It brought three new problems instead.
That last decision, known as soft-fail, hollows out the whole mechanism. An attacker who can present a stolen certificate is, by definition, in a position to interfere with network traffic, so they simply block the OCSP request and the browser proceeds. Adam Langley of Google summarised it memorably: soft-fail revocation checking is like a seatbelt that snaps in a crash.
OCSP stapling moves the work to the server. The server fetches a signed, time-limited OCSP response from the authority and attaches it to the TLS handshake, which removes the latency and the privacy leak. It is one configuration line in nginx, Apache, or IIS, and it should be enabled wherever OCSP is still in use.
Stapling alone does not fix soft-fail, because a client that receives no staple still proceeds. The Must-Staple extension in RFC 7633 marks a certificate as requiring a staple, so a client that supports it will refuse the connection without one. Adoption has stayed low, because a failure in the stapling pipeline then becomes an outage.
Browsers largely stopped relying on either protocol and built their own distribution channels. Chrome ships CRLSets, a curated list of high-priority revocations pushed with the browser. Firefox uses CRLite, a compressed filter structure that encodes the full revocation state of the web in a few megabytes and updates several times a day.
The more decisive answer is to shorten certificate lifetimes until revocation barely matters. Let’s Encrypt turned off its OCSP responders in 2025 after announcing the plan the previous year, relying on CRLs and on 90-day certificates instead. With the CA/Browser Forum reducing the maximum public certificate lifetime to 47 days by 2029, a compromised certificate expires on its own faster than most revocation systems would have propagated the news. That shift is what makes ACME automation a requirement rather than a convenience.
Private PKI has the same problem in a smaller room, and often handles it worse. Typical findings on internal assessments include CRL distribution points that reference a decommissioned server, delta CRLs that stopped publishing, expired CRLs that cause every validating client to fail at once, and VPN or 802.1X configurations that skip revocation checking entirely, so a certificate issued to a laptop that was lost two years ago still authenticates.
If you operate a private certificate authority, publish CRLs to a highly available location, monitor their freshness as you would any other critical service, and decide deliberately whether clients hard-fail. For machine identities, short-lived certificates issued on demand remove the question. We review these settings as part of our infrastructure and architecture review services, and an expired internal CRL is one of the more common causes of a sudden, confusing authentication outage.
Partly. Chrome and Edge consult CRLSets, Firefox uses CRLite, and all of them check revocation properly for EV certificates and for roots and intermediates. Routine OCSP checks on leaf certificates are either skipped or soft-fail, so a blocked responder does not stop the connection. Do not assume a revoked certificate will be rejected by every client.
Yes, where your certificate authority still runs responders. It removes the extra round trip for users, stops the authority from learning who visits your site, and makes status information available even when the responder is briefly unavailable. Monitor the staple, because a server caching a stale or failed response can cause errors for strict clients.
Because expiry is reliable and revocation is not. A client always honours an expiry date, with no network call, no privacy leak, and no soft-fail decision. Reducing lifetimes to weeks limits the usefulness of a stolen key without new infrastructure, which the industry judged a better trade than trying to make revocation checking universal.
Request revocation from the issuing authority immediately, generate a new key pair, and reissue rather than reusing the key. Assume revocation will not reach every client, so also rotate anything the key protected, review logs for use of the certificate, and check Certificate Transparency for other certificates on the same key.
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.