What Is Greylisting?
TL;DR
Greylisting is an anti-spam technique where a receiving server temporarily rejects the first delivery attempt from an unfamiliar sender and accepts the same message when the sender retries. It works because a real mail server queues and retries a temporary failure, and much bulk spam software does not.
How does greylisting work?
The receiver records a tuple identifying the attempt and returns a temporary failure. RFC 6647, the applicability statement for the practice, recommends basing that tuple on the sending IP address, the envelope sender and the first envelope recipient, and leaves the exact 4yz reply code open, noting that 421 and 450 are the reasonable choices.
When a retry arrives matching the same tuple after the configured delay, the message is accepted and the sender is usually remembered, so later mail passes straight through. The cost is a one-off delay on first contact, which is the trade the receiver is accepting deliberately.
What should you do when mail is greylisted?
Nothing. The queue on the sending server handles it, and the message is delivered on the retry without any intervention. The mistake worth avoiding is treating the deferral as a failure and sending the message again as a new send, which produces a duplicate for the recipient and a second first-contact tuple for the receiver.
What is worth doing is reading the pattern. Occasional greylisting on first contact is normal and self-correcting; deferrals that persist across retries at the same provider are a rate or reputation problem being reported through the same 4yz channel.
Frequently asked questions
No. It is a temporary 4yz reply, so the message stays queued and is retried, which is the opposite of a bounce. A sending tool that files greylisting as a bounce will suppress addresses that were never invalid, so check how yours classifies temporary failures.