TL;DR: 552 is either your message being too big or the recipient's mailbox being full, and SendGrid does not retry these, so handle them deliberately. Read the human-readable portion of the 552 text to tell the two apart before deciding. Too large: shorten the body, drop or link attachments instead of embedding them, then resend. Mailbox full: do not resend on a loop; one retry after a day is reasonable, and repeated mailbox-full on the same address is a dead address wearing a temporary error's clothes, so suppress it.

What it looks like:

```text
Symptom: 552, either 'message larger than the current system limit' or
'recipient mailbox full'.
```

Steps:

1. Read the human-readable portion of the 552 text. Success: you know which of the two cases it is.
2. Message too large: shorten the body and drop or link attachments instead of embedding them, then resend. Success: the resend goes through.
3. Mailbox full: wait a day, retry once. Success: delivered, or still full.
4. Track 552s per recipient. Repeated mailbox-full on the same address means suppress it. Success: the dead address stops costing retries.

When to use: 552 responses from SendGrid SMTP sends.

When not to use: 550 rejections (different cause), deferrals (those ARE retried), or the v3 API (the equivalent there is 413 content too large).

Compatibility: SendGrid SMTP; v3 API equivalent is 413.

Variants:
- sendgrid 552 message too large
- sendgrid recipient mailbox full
- sendgrid 552 not retried

Root cause: 552 is permanent for that attempt. It is not a deferral and will not be retried for you, which is why the handling has to be deliberate instead of fire-and-forget.

Edge cases:
- The v3 API equivalent is 413 content too large. Same root cause, same fix.
- A full mailbox is usually an abandoned address. One retry is courtesy; a loop is list rot.