Article

Office 365 NDR Bounce Codes: Full List and What to Do Next

By Unlimited Verifier Team ·

Diagram showing how the first digit of an Office 365 NDR bounce code such as 550 5.1.10 splits failures into permanent 5.x.x and temporary 4.x.x outcomes.

Summary

An Office 365 NDR bounce code is the SMTP reply and enhanced status code Exchange Online returns to explain a failed delivery, such as 550 5.1.10. The first digit of the enhanced code shows whether the failure is permanent (5.x.x, remove the address) or temporary (4.x.x, retry later).

An Office 365 NDR bounce code is the SMTP reply and enhanced status code, such as 550 5.1.10 or 550 5.4.1, that Exchange Online returns in a non-delivery report (NDR) to explain why a message failed. The first digit of the enhanced code tells you whether to remove the address (5.x.x, permanent) or retry later (4.x.x, temporary).

Many teams read an NDR as "this address failed" and move on. The code actually answers three different questions: is the address dead, is the recipient refusing your mail, or is your own sending setup the problem? Each needs a different fix, and mixing them up is how lists get cleaned badly while bounce rates stay high.

For the wider system, see our pillar on email bounce codes. This page focuses on the codes Microsoft 365 and Exchange Online return.

Quick answer

What do Office 365 NDR bounce codes actually mean?

Each NDR carries two codes. The three-digit SMTP reply (such as 550) is defined in RFC 5321, the core SMTP specification. The enhanced status code (such as 5.1.10) is defined in RFC 3463 and has three parts:

Exchange Online adds a text tag after the number. The authoritative text sits on the Diagnostic-Code line that RFC 3464 defines for delivery status notifications. A typical line looks like this (wording varies by scenario):

Diagnostic-Code: smtp; 550 5.1.10 RESOLVER.ADR.RecipientNotFound; Recipient not found by SMTP address lookup

Keep the tag in your export. The number tells you the category of failure. The tag often tells you which part of Exchange refused the message, which is what you need when a spike doesn't match the table below. For a complete reference, see our enhanced SMTP status codes list.

Classification has two costs. Treat a temporary failure as permanent and you delete reachable contacts. Treat a permanent failure as temporary and dead addresses stay on the list, inflating your bounce rate on every future send.

Bottom line: The class digit decides remove versus retry. The subject digit tells you whether the problem is the address, the recipient's policy, or your own sending.

Which NDR codes will you see most often?

A small set of codes accounts for most Exchange Online bounces that many senders see. The table covers nine codes, with the action we recommend for each. Microsoft changes wording over time, so the raw NDR text is the final authority.

Code What it usually means Class Recommended action
5.1.10 Recipient not found in the directory (RESOLVER.ADR.RecipientNotFound) Permanent Suppress; check the source data for typos or mapping errors
5.1.1 Bad destination mailbox address (the generic RFC 3463 form) Permanent Suppress; check the text tag for the system that refused it
5.4.1 Recipient address rejected, access denied; in tenants using directory-based edge blocking, this usually means the address isn't in the directory Permanent Suppress; if all 5.4.1 bounces share one domain, check that domain's policy first
5.7.1 Delivery not authorized, message refused; the recipient limits who can send to them Permanent for you; address may be valid Hold; find out which restriction applies
5.7.509 Sending domain fails DMARC and the domain's policy is reject Permanent until fixed Fix SPF/DKIM alignment before the next send
5.7.515 Sending domain doesn't meet Microsoft's minimum authentication requirements for high-volume senders Permanent until fixed Fix authentication; pause high-volume sends
5.7.606 Sending IP is banned by the recipient's provider Blocked until the IP clears Check IP reputation and blocklists; stop sending from that IP
5.2.2 Mailbox full Temporary in practice Retry; suppress after repeated failures
4.7.0 Temporary deferral; the server asks you to try again later Temporary Retry with slower pacing

Two patterns cause most confusion. First, 5.4.1 and 5.1.10 can both mean the address doesn't exist, but 5.4.1 can also come from access policies. Second, a sudden spike of 5.1.10 right after an import usually points to a mapping problem, such as a CSV column shifted so names landed in the email field, rather than to a list full of dead addresses.

5.7.1 has a sibling. 554 5.7.1 is the same kind of policy refusal from other servers. For 5.1.1 specifically, see the 550 5.1.1 user unknown guide, and for a wider set of fixes, see most common bounce codes and fixes.

Bottom line: Three of these codes describe the address, three describe your sending setup, one describes the recipient's policy, and two are temporary.

How should you classify each NDR code as hard or soft?

Start with the class digit, then override it for policy and setup codes, where the address itself may be fine. The class digit is a standards decision. The override is a practical one.

Sources disagree on mailbox-full failures. RFC 5321 lists 552 among the permanent (5yz) replies, but many senders retry 5.2.2 because full mailboxes often free up. We recommend treating 5.2.2 as soft and suppressing after three failed attempts. That number is our operating rule, not a published standard. For the 4xx and 5xx split, see 4xx vs 5xx SMTP errors.

A hard-bounce count is not an invalid-address count. A single batch can contain 5.7.515 bounces, which are a setup failure, and 5.1.10 bounces, which are dead addresses. Only the second group belongs on the removal list.

Bottom line: The bounce count can't tell you what to remove. The code can, because a 5.7.515 spike and a 5.1.10 spike of the same size need opposite actions.

How do you triage a bounce spike, step by step?

Run this sequence whenever a campaign to Microsoft 365 recipients produces unexpected bounces. The order matters: fix the setup before you touch the list, because cleaning addresses won't stop an authentication failure.

  1. Export 14 days of NDR data around the spike. Include recipient, send time and the full Diagnostic-Code text. Temporary failures can be retried over several days, so a two-week window is our working default.
  2. Extract the enhanced code from every row. Don't rely on your ESP's category label, which can group several different codes under one heading. The tutorial below shows a script.
  3. Count by code, not total. The count per code tells you which fix matters most.
  4. Sort into three buckets: remove now (addressing codes), fix first (authentication, reputation and policy codes), and retry (temporary codes).
  5. Fix the setup before the list. Confirm SPF, DKIM and DMARC alignment and check blocklists. An authentication failure will reject your next send even if the list is perfect.
  6. Verify what remains before the next send. This turns your removal list into a list you can defend.

Bottom line: The code count sets the order. Setup codes come first, then the list.

Where should you collect NDR codes from?

No single source sees every bounce, so choose based on where your mail leaves: Exchange Online directly, an ESP, or both.

Source What you get Limitation
Exchange admin center message trace Delivery status for mail your tenant sent The standard lookback is short; Microsoft documents about 10 days, with longer historical searches available separately. Export promptly.
ESP bounce log or webhook Per-recipient events at high volume Often shows the ESP's category rather than the raw Microsoft code. Ask for the diagnostic text field.
Return-path (bounce) mailbox Full NDR messages, including Diagnostic-Code and text tags Needs parsing, and you own the pipeline. If raw bounces are piling up, see why you keep getting MAILER-DAEMON bounce messages.
Pre-send verification Address-level results before you send Prevents dead addresses from bouncing, but can't explain why a provider refused you

For most teams, the ESP log provides scale, the raw NDR text provides cause for the top codes, and pre-send verification prevents the next wave. If you're weighing vendors, our overview of the best email verification tools and buyer's checklist for choosing a verifier explain what to compare. Judge each option on whether its output helps you act on these codes, not only on email verification pricing.

Bottom line: Use the ESP log for scale, the raw NDR text for cause, and pre-send verification for prevention.

What does a 40,000-contact cleanup look like in practice?

This scenario is illustrative and the counts are hypothetical. The plan limits and arithmetic are real. A B2B company mailed 40,000 contacts, many on Microsoft 365 domains, and received 1,200 bounces, a 3.0% overall rate (1,200 ÷ 40,000).

Code Count Share of bounces Action
5.1.10 640 53.3% Suppress
5.4.1 260 21.7% Suppress
5.7.1 110 9.2% Hold and investigate
5.7.515 90 7.5% Fix authentication; hold
5.2.2 50 4.2% Retry; suppress after 3 failures
4.x 50 4.2% Retry; suppress after 3 failures
Total 1,200 100%

Shares are rounded to one decimal place.

Reading the numbers. The 900 bounces from 5.1.10 and 5.4.1 are dead addresses, so they come out now. The 90 authentication bounces are the more urgent problem. They reflect a setup failure that will hit each future send to those recipients until it's fixed, so fixing it protects far more than 90 contacts.

Cleaning:

  1. Suppress 900 addresses. 40,000 − 900 = 39,100 remain.
  2. Hold 200 addresses (110 for 5.7.1 and 90 for 5.7.515) out of first sends until the restriction is understood and authentication is confirmed.
  3. Keep 100 retry contacts (50 for 5.2.2 and 50 for 4.x) in the list, but suppress any that fail three times in a row. This requires per-address history, which the tutorial covers.
  4. Split for the Pro plan's cap. Pro allows lists of up to 25,000 emails, so 39,100 becomes 25,000 + 14,100 across two uploads.
  5. Check whether the Free plan works. Free allows 25 lists of up to 150 emails a month, or at most 3,750 addresses. 39,100 ÷ 150 = 260.7, so you'd need 261 lists. Free can't handle this list.
  6. Use deep verification where it matters. Standard verification is unlimited on every plan, so run all 39,100 addresses through it. Send deep checks to the 200 held addresses and flagged domains, which uses a small share of Pro's 25,000 monthly deep verifications.

Hypothetical results:

Bucket Count Share
Valid 35,600 91.1%
Risky 2,900 7.4%
Invalid 600 1.5%
Total 39,100 100%

The 600 Invalid addresses never bounced in this send. That's the reason to verify the whole list: they would have bounced on the next one.

Cost: One Pro month at $29 covers this run. Spread over 39,100 addresses, that's $0.74 per 1,000 ($29 ÷ 39,100 × 1,000). On annual billing ($25 per month), the same arithmetic gives $0.64 per 1,000, but you commit to twelve months. Your real cost depends on how many months you need.

Re-entry: Send first to Valid addresses that engaged in the last 90 days. Keep Risky addresses out until the main list shows stable results, then test a small batch of catch-all addresses. Hold the 200 contacts until their codes are resolved. Pause the ramp if any day's bounce rate goes above 1%.

Bottom line: The largest code (5.1.10) isn't the most urgent one. The 90 authentication bounces will repeat on each send until they're fixed, which is why they outrank 640 dead addresses.

How do you fix each code family?

Each family has a different owner. Addressing codes are yours to remove. Policy codes belong to the recipient. Setup codes are yours to fix at the source.

Addressing codes (5.1.1, 5.1.10, 5.4.1). The address is the problem. Remove it, then check the source for typos, decommissioned domains and mapping errors. Don't re-add it through a form or import without verifying it first. The lasting fix is validation at the point of entry, covered in real-time validation at signup. For 5.4.1 specifically, see 550 5.4.1 recipient address rejected.

Recipient policy (5.7.1). The address works, but the recipient's system accepts mail only from approved senders. You can't change that from your side. Check whether your sending domain or IP is on their allow list, and whether the recipient is a distribution list with sender restrictions. Keep these contacts out of bulk sends until you understand the restriction.

Authentication (5.7.509, 5.7.515). These cost the most because they repeat. Confirm that SPF includes every service that sends on your behalf, that DKIM signs with a domain you control, and that DMARC aligns with at least one of them. Microsoft has tightened authentication requirements for high-volume senders in recent years, so check the current rules in Microsoft's Exchange Online documentation before you resume volume.

Reputation (5.7.606). A banned IP is an infrastructure problem, not a list problem. Check the IP against major blocklists and look for shared-IP neighbours sending spam. Move to a dedicated IP only if your volume justifies the cost. Stop sending from the affected IP until it clears.

Temporary codes (4.x, 5.2.2). Retry with slower pacing and suppress only after repeated failures. A 4.x deferral often means you're sending faster than the recipient's server accepts, so slowing down usually clears it. For details, see 421 4.7.0 try again later and SMTP 450 requested mail action not taken, or the list of common soft bounce reasons.

Bottom line: Only addressing codes are fixed by removing the address. Setup and policy codes are fixed on your side or with the recipient, and removing those addresses only hides the cause.

How do you verify the remaining list? (hands-on tutorial)

Steps 1–3 work on your bounce export. Steps 4–8 run the remaining list through verification. You'll need a free sign up for an account, and the Free plan works for testing.

Step 1: Export your bounce data. Save one row per NDR, with recipient address, timestamp and the full Diagnostic-Code text. Name the text column diagnostic.

Step 2: Extract and count the enhanced codes. Run this script on the export:

import csv
import re
from collections import Counter

# Matches codes like 5.1.10 or 4.7.0, but not IP addresses or decimals inside text
ENHANCED = re.compile(r'(?<![\d.])[245]\.\d{1,3}\.\d{1,3}(?![\d.])')
counts = Counter()

with open('ndr_export.csv', newline='', encoding='utf-8') as f:
    for row in csv.DictReader(f):
        m = ENHANCED.search(row['diagnostic'] or '')
        if m:
            counts[m.group(0)] += 1

for code, n in counts.most_common():
    print(code, n)

Spot-check ten rows before you trust the output. The script counts NDRs, not addresses. For the three-failure rule, group rows by address, sort by time, and count consecutive soft failures. A pivot table in any spreadsheet can do this.

Step 3: Build the suppression list. Include every address with 5.1.1, 5.1.10 or 5.4.1, plus any address with three consecutive soft failures. Keep this file. It's your evidence if you need to explain the cleanup to an ESP or a client.

Step 4: Create the verification file. Remove the suppressed addresses and save the rest as a CSV with an email column. Hold the 5.7.x addresses in a separate file. Upload accepts CSV or Excel files up to 10MB, and the email column is detected automatically. Lists over 25,000 rows need splitting on Pro, and Free handles lists of up to 150 emails.

Step 5: Upload and verify. Upload and verify a list and run the file. Each address is checked for syntax, domain, MX records and the SMTP mailbox, then sorted into Valid, Risky or Invalid. Catch-all domains, disposable addresses and role addresses are flagged. The verifier checks addresses, not NDR text, so the codes decide what to remove and the verifier decides what's safe to send.

Step 6: Handle Risky results. Risky includes catch-all domains, which accept every address, so the specific mailbox can't be confirmed. Catch-all email verification explains how these results are handled. Keep Risky addresses out of your first sends.

Step 7: Export before results expire. Download as CSV, Excel or JSON. Results are kept for 24 hours on Free and 30 days on Pro.

Step 8: Re-import only what's safe. There are no native ESP, CRM or Zapier integrations, so import the Valid CSV into your ESP as a new segment, and add the Invalid addresses to its suppression list. To keep the list clean between uploads, follow an email list maintenance routine.

Automate the next run. To test a handful of addresses first, the free email checker runs single-address checks. If you're checking addresses from your own code, the email verification API accepts single checks and bulk batches of up to 10,000 emails per request. API access is included on Pro, with 100,000 emails a month, and not on Free. For the wider workflow patterns, see email verification API and automation.

Bottom line: Extraction tells you what to remove. Verification tells you what is safe to send. Neither step replaces the other.

What mistakes do teams make with NDR codes?

Bottom line: Most mistakes come from reading the bounce count instead of the code. The code is where the fix starts.

How we compared (and limitations)

This guide draws on public documentation: RFC 5321, RFC 3463, RFC 3464, Microsoft's Exchange Online documentation, and Google's email sender guidelines for general bulk-sender expectations. It is not based on Microsoft's internal rules.

We haven't run controlled tests against Microsoft's delivery systems, and we haven't confirmed that every code and tag described here appears in every tenant. Microsoft changes codes and wording over time, so check raw NDR text against current documentation.

The 99.5% accuracy figure for Unlimited Verifier's deep verification is the company's stated figure, not an independent benchmark. The worked example uses hypothetical counts. The plan limits and arithmetic are real. Our thresholds (under 2% total bounces, under 1% on re-entry, and three failed attempts before suppressing a soft bounce) are operating rules, not industry standards.

Bottom line: Trust the raw NDR text over any summary, including this one.

Common NDR codes and the action to take

Code Class Likely cause Action
550 5.1.10 Permanent Recipient not found in Exchange Online Suppress after one failure, verify the address
550 5.1.1 Permanent Mailbox does not exist Suppress after one failure
550 5.7.1 Permanent Recipient restriction or policy block Do not auto-delete, review diagnostic text
550 5.7.515 Permanent Sender authentication failure Fix SPF, DKIM, and DMARC before re-sending
4.x.x (temporary) Temporary Full mailbox, throttling, or delay Retry, suppress after three consecutive failures

Frequently asked questions

What does a 550 5.1.10 bounce mean in Office 365?

A 550 5.1.10 bounce means Exchange Online could not find the recipient address in its directory, so the recipient does not exist there. It is a permanent failure. Remove the address from your list after one failure, and first check for typos or a retired mailbox.

What is the difference between 5.1.1 and 5.1.10?

Both are permanent addressing failures, but they describe the address differently. 5.1.1 means the mailbox does not exist at the destination, while 5.1.10 means Exchange Online's recipient lookup found no matching recipient. Either way, suppress the address and confirm it before you ever re-send to it.

What does a 5.7.1 NDR code mean?

A 5.7.1 code is a security or policy rejection, and it often means the recipient restricts which senders can email them, or your message was blocked by a policy. The address may still be valid. Don't delete it automatically. Check the diagnostic text and consider contacting the recipient through another channel.

What do 5.7.509 and 5.7.515 mean?

These codes indicate a sender authentication problem, usually related to SPF, DKIM, or DMARC alignment for your sending domain. The address is not the issue. Fix your DNS authentication records first, then re-send. Removing these recipients would wrongly shrink a healthy list.

Should I remove addresses that return a 4.x.x code?

Not right away. A 4.x.x code is a temporary failure, such as a full mailbox or a throttling response, and the server will retry later. A common operating rule is to suppress an address only after three consecutive temporary failures, which avoids deleting reachable contacts.

What is the RESOLVER.ADR tag in an NDR?

The RESOLVER.ADR tag is part of the Exchange Online diagnostic text that follows the enhanced status code. It indicates the failure came from the address resolution stage, which is consistent with recipients not being found. Keep the full tag in your bounce exports so you can group failures by cause.

Where can I find the official definitions of these codes?

The three-digit SMTP reply codes come from RFC 5321, and the enhanced status code format comes from RFC 3463. The Diagnostic-Code field in delivery notifications is defined in RFC 3464. Microsoft's Exchange Online documentation adds service-specific tags and explanations for its own NDRs.

How do I tell whether my bounces come from bad addresses or my own setup?

Group your bounces by enhanced code before you clean anything. Addressing codes such as 5.1.x point to bad addresses. Authentication codes such as 5.7.509 and 5.7.515 point to your sending setup. Policy codes such as 5.7.1 point to recipient restrictions. Each group needs a different fix.

Cut Bounces and Protect Your Sender Reputation With Unlimited Email Verification

Every invalid address you send to hurts your sender reputation and your inbox placement. Verifying your list before each send means fewer bounces, healthier deliverability, and a predictable cost. With Unlimited Verifier, standard verification is unlimited on every plan, with no per-email fees.

Unlimited Verifier checks syntax, domain, MX records and the SMTP mailbox. It also detects catch-all domains and flags disposable and role-based addresses. Export a CSV from your email platform or CRM, upload and verify the list, then re-import the clean results. You can also automate checks with the email verification API.

Free Pro
Price $0 $29/month ($25/month billed annually)
Lists 25 a month, up to 150 emails each Unlimited, up to 25,000 emails each
Deep verifications 750 a month 25,000 a month
API access No Yes (100,000 emails a month)
Result retention 24 hours 30 days

Why teams choose Unlimited Verifier

Unlimited Verifier states 99.5% accuracy for deep verification.

Create a free account and verify your next list before you hit send.

Want to test a single address first? Try the free email checker.