Guide · 8 min read
How to send email through a SOCKS5 proxy — and why most mail clients can't
Routing a browser through a proxy takes one setting. Routing email through one is harder than it looks: there are two protocols, most clients have one proxy setting for everything, and a send can quietly leave by a direct connection while every setting says otherwise. Here is what it actually takes.
Why route email through a proxy at all
When a desktop mail client sends a message, it connects to your provider's SMTP server, and that server usually
records where the connection came from. The record goes into the message itself, as a Received:
header, and every recipient keeps it:
Received: from WS-OFFICE (unknown [203.0.113.42])
by smtp.example-provider.com with ESMTPSA id 4F1A2C
Fri, 13 Nov 2026 09:07:39 +0200
That address locates the office or home the message was written in, and it is the same on every message that machine sends. If you run mail for several companies from one desk, the same address in each company's mail is the thread that connects them.
A SOCKS5 proxy changes the origin of the connection: the provider sees the proxy's exit address instead of yours, and that is what ends up in the header. You can see what your own messages carry with our free header checker.
Two connections, not one
Email is two protocols, and both have to go through the proxy:
- IMAP (usually port 993) — reading mail. It shows your provider your address every time the client checks for new mail, often every few minutes.
- SMTP (port 587 or 465) — sending. This is the one that is written into the message.
Proxying only one of them is the common half-measure. If reading goes direct, your provider's sign-in log shows your real address next to the proxy's; if sending goes direct, the recipient sees it.
What the usual mail clients offer
| Client | Proxy support | Per mailbox? |
|---|---|---|
| Thunderbird | One SOCKS proxy in the network settings, for everything the app does | No — every account shares it |
| Outlook (desktop) | Follows the operating system's proxy settings; no SOCKS setting of its own | No |
| Apple Mail | Follows the system proxy | No |
| System VPN | Moves the whole machine to one exit | No — one address for all mail |
| Per-app proxy tools | Route one application through one proxy | No — per app, not per account |
All of these give you one exit for every mailbox. That hides your own address, but it still puts the same address in every company's mail — which is exactly the link you may be trying to avoid.
The quiet leak: a send that skips the proxy
The dangerous failure is not an error. It is a send that works, while the message leaves by a direct connection. It happens in a few ordinary ways:
- Silent fallback. The proxy is down, the software retries without it, and the message goes out anyway. Nothing tells you.
- Half-proxied sessions. Some software routes the first connection through the proxy and opens a second, direct one for the actual SMTP conversation. Every log says "proxy"; the recipient's copy says otherwise.
- Only part of the app is proxied. The system or app proxy covers reading but not sending, or the other way round.
The only defence is a design that fails closed: if the route is unreachable, nothing is sent and
nothing is fetched, and the operator is told why. Then check the result rather than trusting the setting — send a
message to yourself and look at its first Received: line.
Doing it per mailbox
If every company mailbox should have its own exit, the proxy has to belong to the mailbox, not to the application. What that requires:
- A route per mailbox — its own SOCKS5 endpoint, used for both IMAP and SMTP.
- No fallback — a dead route stops that mailbox, and only that mailbox.
- The sending identity fixed — the message is written from the mailbox that owns the route, so it cannot leave from the wrong one.
- Proof in the headers — after sending, the exit address in the message is the route's, and you can check it.
How SecureMail does it
SecureMail is a mail client built around exactly this. Each mailbox has its own SOCKS5 route; reading and sending both go through it; there is no direct fallback anywhere; and the client shows each message's raw source with the address-bearing headers marked. You bring the proxies — we route through them, but never sell or run them. Your mail stays on your mail server, and we keep no copy of it.
See what your own mail reveals
Paste the headers of a message you sent and see which address it carries — or ask us for a walkthrough.
Check my email headers Book a walkthrough