You can verify an email address without sending by holding the first half of an SMTP conversation with the recipient's mail server: look up the MX record, connect on port 25, say EHLO, give a MAIL FROM, ask RCPT TO:<address>, read the reply, then QUIT before any message is sent. A 550 5.1.1 reply means the mailbox does not exist; a 250 only means something if the same server rejects a random made-up address on that domain. The check fails from most home and cloud connections because outbound port 25 is blocked, and it proves nothing on catch-all domains.
This guide shows the full conversation on a fictional domain, explains each reply, and covers when the check is not worth running. For what each final verdict means, see the pillar guide, email verification results explained.
Before you start: what this check can and cannot do
| It can tell you | It cannot tell you |
|---|---|
| The domain does not accept mail (null MX, no mail host) | Whether your real message will be accepted |
| The server rejects this mailbox name now | Whether the mail will reach the inbox or spam |
| The server accepts every address (catch-all) | Which addresses on a catch-all domain are real |
| The server is deferring or blocking you | Whether the person still reads that mailbox |
Step 1: find the mail server
$ dig +short MX oakline-joinery.example
10 mx1.mailhost.example.
20 mx2.mailhost.example.
Use the host with the lowest preference number (mx1). Two special cases:
- Null MX. A single record
0 .is the RFC 7505 signal that the domain accepts no mail. Stop: every address on it is invalid. - No MX. RFC 5321 section 5.1 says that "if an empty list of MXs is returned, the address is treated as if it was associated with an implicit MX RR", so the domain's own A or AAAA record is the mail host. Many such domains have no mail service at all; the connection will often be refused.
Step 2: connect on port 25
$ nc mx1.mailhost.example 25
220 mx1.mailhost.example ESMTP ready
telnet mx1.mailhost.example 25 works the same way. Port 25 is the server-to-server SMTP port. Ports 587 and 465 are for authenticated submission from your own mail client to your own provider; they will not let you ask a third party's server about its users.
Why port 25 matters
If the command hangs and then times out, nothing is wrong with the recipient. Your network blocks outbound port 25. Most residential broadband providers do this to stop infected PCs sending spam directly, many offices do the same, and most cloud platforms block or restrict outbound port 25 on new accounts by default. The result is that a home-built verifier reports everything as unknown.
You need a network that permits outbound port 25. Even then, receiving servers judge the connecting IP: a residential IP or one without reverse DNS will often be greeted with a rejection such as 554 5.7.1 Client host rejected before you can ask anything. That is a verdict on you, not on the address.
Step 3: introduce yourself
EHLO verify.oakline-tools.example
250-mx1.mailhost.example Hello verify.oakline-tools.example
250-SIZE 52428800
250-PIPELINING
250-STARTTLS
250 8BITMIME
MAIL FROM:<[email protected]>
250 2.1.0 Sender OK
Three things make this step succeed:
- A real hostname in EHLO whose forward and reverse DNS match the IP you connect from. Many servers reject or score down a bare IP or a made-up name.
- A sender address on a domain you control. Some verifiers use the null sender
<>; some servers refuse it for this purpose, and using another organisation's domain is spoofing. - Patience. RFC 5321 section 4.5.3.2 suggests clients wait up to 5 minutes for a reply to
MAILand toRCPT. Servers that deliberately slow down unknown clients will look like timeouts to a tool that gives up after seconds.
The server may offer STARTTLS. Encrypting the session is good practice and some servers require it before accepting MAIL FROM.
Step 4: ask about the recipient
RCPT TO:<[email protected]>
250 2.1.5 Recipient OK
Or, for an address that does not exist:
RCPT TO:<[email protected]>
550 5.1.1 <[email protected]>: Recipient address rejected: User unknown
Read three parts of every reply: the basic code (250, 550), the enhanced status code (5.1.1) and the text.
| Reply | Meaning | Verdict |
|---|---|---|
250 2.1.5 | Accepted | Provisional valid; run step 5 |
251 | Accepted, will forward | Provisional valid |
252 | "Cannot VRFY user, but will accept message and attempt delivery" (RFC 5321) | Accepted unconfirmed |
450, 451 4.7.1 | Temporary failure, often greylisting | Unknown; retry in 15 to 30 minutes |
452 4.5.3 | Too many recipients this session | Unknown; slow down |
550 5.1.1, 550 5.1.0 | No such mailbox | Invalid |
550 5.4.1 Recipient address rejected: Access denied | Microsoft 365 rejecting an unknown recipient at the edge | Invalid |
553 5.1.3 | Mailbox name not allowed (bad syntax) | Invalid |
550 5.7.1, 554 5.7.1 | Policy rejection of your connection or sender | Unknown; the problem is your checker |
The 5.4.1 row is worth knowing. Microsoft's Directory-Based Edge Blocking rejects messages for invalid recipients at the perimeter with 550 5.4.1 Recipient address rejected: Access denied, so Microsoft 365 domains often give clear answers. Google Workspace normally rejects unknown recipients with a 550 5.1.1 reply too, unless the administrator has set up a catch-all routing rule.
Step 5: the catch-all control
A 250 on its own is weak. Before trusting it, ask the same server, in the same session, about an address that cannot exist:
RCPT TO:<[email protected]>
250 2.1.5 Recipient OK
If the random address is accepted too, the server accepts everything at this stage. The domain is catch-all, and the earlier 250 for sam.reid@ proves nothing. If the random address is rejected with 550 5.1.1, the server is checking its directory, and the earlier 250 is meaningful.
Use a random string long enough that it cannot collide with a real mailbox, and generate a new one per domain.
Step 6: quit without sending
RSET
250 2.0.0 Reset OK
QUIT
221 2.0.0 Bye
Never send DATA. Record the result with the date, the MX host, the reply codes and texts, so a later reader can see why the address was labelled the way it was.
The full annotated session
$ nc mx1.mailhost.example 25
220 mx1.mailhost.example ESMTP ready <- server greeting
EHLO verify.oakline-tools.example <- your real hostname
250-mx1.mailhost.example Hello <- capabilities follow
250 8BITMIME
MAIL FROM:<[email protected]> <- sender you control
250 2.1.0 Sender OK
RCPT TO:<[email protected]> <- the address you are checking
250 2.1.5 Recipient OK <- provisional yes
RCPT TO:<[email protected]> <- catch-all control
550 5.1.1 User unknown <- server checks its directory
RSET <- abandon the transaction
250 2.0.0 Reset OK
QUIT
221 2.0.0 Bye
Verdict: [email protected] is deliverable at the time of the check.
Providers and setups that always accept
Some servers answer 250 to everything, which makes the check uninformative:
- Catch-all routing. The organisation sends all unknown addresses to one mailbox. Google Workspace, Microsoft 365 and most hosting control panels support this.
- Security gateways. Many organisations put a filtering gateway in front of their mailboxes. Whether the gateway checks recipients against a directory at RCPT or accepts everything and decides later is a configuration choice, and from outside you only see the result.
- Accept now, bounce later. RFC 5321 notes that some servers "do not perform recipient verification until after the message text is received" and report failures afterwards as a bounce.
- Hybrid and migration setups. A Microsoft 365 domain configured as internal relay passes unknown recipients on rather than rejecting them at the edge.
We look at the Microsoft 365 and Google Workspace cases in catch-all on Microsoft 365 and Google Workspace.
Rate and etiquette
RFC 5321 section 7.3 records that commands like EXPN were used to "harvest" addresses, which is why most servers disable VRFY and EXPN and watch for the same behaviour with RCPT. Your probe should look nothing like harvesting:
- Check only addresses you have a reason to contact, not every possible name on a domain.
- One connection per domain at a time, with a pause between domains on the same MX host.
- A few recipients per session at most. One real address plus one control is ideal. A
452 4.5.3reply means you asked for too many. - Respect temporary failures. A
4xxreply means come back later, not try again immediately. - Never probe from your sending IP. If your verifier gets listed, you do not want your campaign mail to share the listing.
- Use real identity. A matching EHLO hostname, reverse DNS and a sender address that accepts replies.
- Stop on blocks. A
5.7.xpolicy rejection means stop checking that host, not rotate IPs.
Some providers say this explicitly. Microsoft's Outlook.com policies ask senders not to use namespace mining, which is exactly what bulk recipient probing looks like from the receiving side.
When to just not verify
The SMTP check is not always worth it:
- Addresses published on the company's own contact page. They almost always exist; a syntax and MX check is enough, and the catch-all question does not change your decision much.
- Catch-all domains you already know about. The answer will be the same every time.
- Small lists you will send to by hand. A bounce tells you the same thing with no probing.
- When your network blocks port 25. You will get a list full of unknowns; use a service whose infrastructure allows the check, or skip it.
- When you have no lawful basis to email the address. Verification is not a reason to collect or contact anyone.
If you do send without an SMTP check, keep the volume small and let a bounce limit stop the campaign if the list is worse than you thought.
How Map Extract helps
The email verifier in Map Extract runs this conversation for you from your own PC. It checks syntax and IDNA, looks up MX records with null MX and implicit MX handling, and can run an optional SMTP mailbox probe on port 25 with a random-recipient catch-all control. It needs a probe sender address you control, and records a reason such as probe_sender_required when one is missing. Results are recorded as deliverable, invalid, catch_all, accepted_unconfirmed, domain_valid or unknown, with the reason alongside, rather than a bare yes or no.
Addresses come from the email extractor, which records where each was found and what the inbox is for, so you can skip probing addresses published on a contact page if you choose. In the bulk email sender, settings decide whether catch-all and unknown results are allowed to send.
What it does not do: it cannot open port 25 on a network that blocks it (you will see unknown results), it does not rotate IPs or work around blocks, and no verdict guarantees delivery.
Related guides
- Email verification results explained
- Catch-all on Microsoft 365 and Google Workspace
- MX record lookup
- SMTP error code lookup
Key takeaways
- The check is the first half of an SMTP conversation: MX lookup, connect on 25, EHLO, MAIL FROM, RCPT TO, quit.
550 5.1.1and550 5.4.1are clear invalids;5.7.xrejections are about your checker, not the address.- A
250counts only if a random address on the same domain is rejected. - Port 25 is blocked on most home and cloud networks, which turns every result into unknown.
- Probe slowly, with real identity, only for addresses you have a reason to contact.
Frequently asked questions
Can you check if an email address exists without sending an email?
Often, yes. You look up the domain's mail server, open an SMTP connection on port 25, give a sender address, name the recipient with RCPT TO and read the reply before quitting. A 550 with a 5.1.1 code means the mailbox does not exist. A 250 means the server accepted it, which only counts if the same server rejects a random made-up address on that domain.
Why does telnet to port 25 time out from my computer?
Most home broadband and many office networks block outbound connections to port 25 to stop infected machines sending spam, and most cloud providers block or restrict it by default too. The connection simply never opens. You need a network that permits outbound port 25, ideally a server with a proper hostname and reverse DNS, or a verification service that has one.
Is the SMTP VRFY command still useful for checking emails?
Rarely. RFC 5321 lets administrators disable VRFY and EXPN for security reasons, because they were used to harvest addresses, and most public mail servers either disable them or reply 252, which means the server cannot verify the user but will accept the message. RCPT TO inside a mail transaction is the check that still gives useful answers.
Can an SMTP check get my IP blocklisted?
It can if you behave like a directory harvester: many RCPT commands per connection, many connections per minute, invalid HELO names or probing from the same IP that sends your campaigns. Keep the rate low, use a real hostname with matching reverse DNS, test only addresses you have a genuine reason to contact, and never probe from your outbound mail server.
What does it mean if every address on a domain returns 250?
The server is accepting all recipients at the RCPT stage, which is called catch-all or accept-all. It may deliver everything to a shared mailbox, or a security gateway may accept now and bounce later. Either way the 250 tells you nothing about the specific mailbox. Treat those addresses as unconfirmed and decide on other evidence, such as being published on the company's own site.
Sources
Written by the Map Extract editorial team. Found an error? Email [email protected] and we will check and correct it. How we research guides: editorial policy.