What you're actually defending against

It helps to picture the adversary accurately. Nobody is personally targeting your server; you are facing automation. Botnets continuously scan the internet for port 3389, then run through leaked username/password pairs, dictionary words with number suffixes, and default credentials. When guessing fails, they probe for known unpatched vulnerabilities. The whole pipeline runs without human attention until something answers — and then a human (or ransomware script) takes over.

That picture dictates the defense. You do not need exotic countermeasures; you need to break every link in that automated chain. A port the bots cannot reach, credentials they cannot guess, software without known holes, and logs that alert you when something odd happens anyway. Each best practice below severs one link, and together they leave the automation with nothing to grip. (On RDP Temple plans — Shared $6/mo, Virtual $28/mo, Dedicated $165/mo — you hold full admin access, so every control below is yours to apply.)

Botnet scans + guesses Defense layers firewall scope strong credentials + 2FA patched OS + NLA lockout policy log monitoring RDP server hardened ✗
Fig. 1 — Layered defenses break the automated attack chain before it reaches your server.

The 12-point RDP hardening checklist

Work through these in order. The first six eliminate the vast majority of real-world compromises; the rest turn a hardened server into a maintainable one.

  1. Install updates promptly. Enable automatic Windows updates, or patch on a fixed weekly schedule. Unpatched RDP flaws are the fastest route in after weak passwords.
  2. Require Network Level Authentication (NLA). NLA forces authentication before a full session starts, which blunts denial-of-service and pre-auth exploit attempts.
  3. Use long, unique passwords — 16+ characters, generated by a password manager, never reused anywhere else. Usernames should not be "Administrator" if you can avoid it.
  4. Add two-factor authentication to the Windows logon so a stolen password alone is useless. See our 2FA guide for the options.
  5. Set an account lockout policy so repeated guessing locks the account instead of eventually succeeding. Details in our lockout policy guide.
  6. Scope the firewall rule. Restrict the inbound Remote Desktop rule to only the IPs that need it — your home, office, or VPN range. Our firewall rules guide walks through it.
  7. Consider a VPN in front of RDP so port 3389 is never directly reachable from the internet at all. Covered in securing RDP with a VPN.
  8. Enforce TLS encryption for the RDP security layer so sessions cannot be downgraded to weak encryption. See RDP TLS encryption.
  9. Disable what you don't use. If nobody needs RDP on a machine, turn it off rather than defending an unused door.
  10. Review local users regularly. Remove stale accounts, and make sure every remaining account has a reason to exist and a strong password.
  11. Monitor failed logons. A sudden spike of Event ID 4625 failures is your early-warning system — investigate before it becomes a breach.
  12. Keep tested backups. If the worst happens, clean backups are the difference between a bad day and a catastrophe.

Patching: the unglamorous lifesaver

Every few years a critical RDP vulnerability makes headlines precisely because so many servers sit unpatched. The pattern is always the same: a flaw is disclosed, working exploit code circulates within days, and bots sweep the internet for servers that missed the update. The servers that get hit are not unlucky — they are simply behind.

The fix is boring on purpose. Turn on automatic updates for security patches, or hold a non-negotiable weekly maintenance window. If you run software that is sensitive to reboots, schedule updates for off-hours rather than postponing them indefinitely. A server that reboots once a month for patches is dramatically safer than one with eleven months of uptime and eleven months of known holes.

Accounts: fewer admins, stronger secrets

Every account on the server is a lottery ticket for attackers, so hold as few as possible. Disable or rename the built-in Administrator account, create named admin accounts for actual humans, and give day-to-day work a standard (non-admin) account where practical. Service accounts should have only the rights their job requires and passwords nobody has ever typed by hand.

On the password side, length beats complexity theater. A 20-character passphrase generated by a password manager resists both guessing and cracking far better than an 8-character tangle of symbols that gets reused across sites. And because even the best password can leak, treat two-factor authentication as the companion control, not an optional extra.

Network: shrink what's reachable

The most effective network control is also the simplest: decide exactly which IP addresses legitimately connect, and let the firewall deny everything else. For most people that list is tiny — a home IP, an office range, a VPN subnet. Scoping the Remote Desktop firewall rule to that list (both the TCP and UDP rules) quietly deletes 99% of attack traffic before authentication is even attempted.

Changing the RDP port away from 3389 is sometimes suggested here. Treat it as obscurity, not security: it cuts down log noise from dumb scanners, which is genuinely useful for readability, but a targeted attacker port-scans in seconds. Do it for quieter logs if you like, but never instead of scoping, NLA, and strong credentials.

Monitoring: read the logs before attackers do

Windows logs every failed logon as Event ID 4625 in the Security log. A healthy server shows a trickle of these — typos, mostly. A server under attack shows a flood: hundreds of failures per hour from a handful of IPs, cycling through usernames. Learn what your baseline looks like so the flood is obvious.

You do not need enterprise tooling to benefit. Once a week, open Event Viewer, filter the Security log for 4625, and glance at the volume and source IPs. If the numbers jump, check that your lockout policy and firewall scoping are actually in place — an attack succeeding usually means one of them drifted. Consistent small attention beats heroic incident response every time.

↻ Patch weekly updates reboot if needed 👤 Accounts remove stale users rotate secrets 🔍 Logs failed logons spike = investigate 💾 Backups verify restores off-server copy
Fig. 2 — A weekly routine that keeps every other control honest: patch, accounts, logs, backups.

Backups and recovery

Security guides sometimes treat backups as an afterthought; ransomware made them a front-line control. If an attacker does get in and encrypts the disk, your options are paying criminals or restoring from backup — and only one of those is acceptable. Keep backups automated, keep at least one copy off the server itself (and ideally offline or immutable), and periodically verify that a restore actually works.

Test restores are the part everyone skips. A backup job that has been "succeeding" for a year but has never been restored from is a hope, not a plan. Once a quarter, restore a file or two to a scratch location and confirm the data is intact. Fifteen minutes of verification buys certainty that no dashboard green light can provide.

What each practice stops (and what it costs you)

If you are prioritizing, this table maps each control to the attack it defeats and the effort involved. Start at the top: high impact, low effort first.

PracticeAttack it stopsEffort
Prompt patchingExploit kits targeting known RDP flawsLow — automate it
Firewall scopingInternet-wide scanning and guessingLow — one-time rule
Strong unique passwordsCredential stuffing and dictionary attacksLow — use a manager
Account lockout policySustained brute-force guessingLow — one-time setting
NLA requiredPre-authentication exploits and DoSLow — one checkbox
Two-factor authenticationStolen or leaked passwordsMedium — install + enroll
VPN in front of RDPAll direct internet attacks on 3389Medium — tunnel setup
Log monitoringSlow or novel attacks the rest missMedium — weekly habit
Tested backupsRansomware and destructive breachesMedium — ongoing

A weekly 15-minute security routine

Controls decay — rules get widened "temporarily", accounts accumulate, patches slip. A short weekly routine keeps everything above honest:

  • Confirm updates installed. Check Windows Update history for the latest security patches and any pending reboot.
  • Glance at failed logons. Filter the Security log for Event ID 4625; investigate any sharp rise in volume or new source IPs.
  • Review the firewall scope. Make sure the Remote Desktop rule still lists only your approved IPs — no "temporary" widenings left behind.
  • Check user accounts. Disable anything you don't recognize and confirm ex-team-members or old service accounts are gone.
  • Verify the backup ran. Confirm the latest backup completed and spot-check that the off-server copy is reachable.

Remember: security is a habit, not a product. No checklist protects a server whose owner set it once in 2024 and never looked again. The practices above are deliberately low-effort so you will actually keep doing them — and done consistently, they put your RDP server ahead of the vast majority of targets the bots are scanning.

Ready for your own RDP server?

Full admin access, dedicated IP, 24/7 support and a 3-day money-back guarantee on every plan. Pay by credit/debit card, bank transfer, Bitcoin or altcoins.

Shared Servers$6/mo
Dedicated Server$165/mo
Get Started

3-day money-back guarantee · Instant setup · 30+ server locations · 99% uptime SLA

RDP Security Best Practices: FAQ

Keeping the server patched. Unpatched RDP vulnerabilities are exploited by automated bots within days of disclosure, and no password policy helps against a flaw in the protocol handler itself. Enable automatic security updates or hold a fixed weekly patching window — everything else builds on this foundation.

Changing the port from 3389 reduces log noise from casual scanners, which makes your logs easier to read. But it is obscurity, not security: a targeted attacker finds the real port with a quick scan. Change it if you like quieter logs, but never treat it as a substitute for firewall scoping, strong credentials, and NLA.

A quick weekly review is enough for most servers: filter the Security log for failed logons (Event ID 4625) and look for spikes in volume or unfamiliar source IPs. Investigate immediately if failures jump sharply — that pattern means an active guessing campaign, and you want your lockout policy and firewall scoping confirmed while it is happening.

Not strictly, but it is one of the strongest single controls available. A VPN plus a scoped firewall rule removes your RDP port from the public internet entirely, which neutralizes scanning and brute-force attacks at a stroke. If a VPN does not fit your setup, strict firewall scoping to known IPs achieves much of the same effect.

It is strongly recommended. Passwords leak through breaches, phishing, and reuse, and 2FA means a stolen password alone cannot open a session. With full administrator access on your server you can install and configure MFA for the Windows logon yourself — it is one of the highest-value upgrades you can make.

Disconnect it from the network first to stop further damage, then investigate from a clean angle: review logons, scheduled tasks, and new accounts. Assume credentials are burned and rotate them everywhere they were reused. In most cases the safest recovery is rebuilding from a known-clean image and restoring data from tested backups, rather than trying to clean a compromised system in place.