Quick answer: Enable an account lockout policy (lock the account after 5 bad attempts for 30 minutes), enforce long unique passwords on every account, rename or disable the default Administrator account, and restrict RDP to known IP addresses in the firewall wherever possible. Add Network Level Authentication so attackers must authenticate before getting a session. Each layer multiplies the attacker's work.

What a brute force attack actually looks like

Forget the Hollywood image of a hacker typing furiously. Real RDP brute forcing is industrial: botnets of compromised machines each try a few hundred common passwords against thousands of servers, moving on the instant an account locks. The most common usernames tried are Administrator, admin, and user; the passwords are lists with entries like "Password123," seasonal variants, and keyboard walks. A server with the default Administrator account and a weak password can fall in a single night.

The economics favor the attacker only when the target is soft. Every defense you add — a lockout policy, a non-guessable username, a nonstandard port, an IP allowlist — does not need to be perfect; it needs to make your server more expensive to crack than the ten thousand others the botnet is scanning. Brute force protection is about raising cost, layer by layer, until the bots move on.

One uncomfortable truth first: if RDP is reachable from the entire internet, you will be attacked, continuously, forever. That is the background radiation of the internet. The goal is not zero attempts — it is zero successful attempts, plus logs that prove it.

Botnet password lists 24/7 attempts IP allowlist Account lockout Strong passwords Your server port 3389 holds firm Layers multiply the attacker's cost until the bots move on.
Fig. 1 — Brute force is constant; layered defenses make it futile.

Read the warning signs in Event Viewer

Before hardening, look at what is already happening. On the server, open Event Viewer > Windows Logs > Security and filter for event ID 4625 — failed logon attempts. On an internet-exposed server you will likely see bursts of them, often dozens per hour, with usernames like Administrator that do not match your accounts. Each 4625 entry shows the source IP, so you can see whether the attempts come from a handful of persistent hosts or a rotating botnet.

Also check TerminalServices-RemoteConnectionManager under Applications and Services Logs > Microsoft > Windows for connection-level events. A healthy pattern is a small number of 4625s from your own mistyped passwords; an attack pattern is hundreds from foreign IPs against accounts you never created. If you see the latter and have no lockout policy yet, treat hardening as urgent — the bots have already found you.

Make this a habit, not a one-time check. A weekly five-minute glance at the Security log tells you whether your defenses are working: attempts should continue (that never stops), but lockouts and failures should be the whole story, with zero successful logons you do not recognize (event ID 4624, logon type 10 for RDP).

Layer 1: account lockout policy

The account lockout policy is the single highest-value defense: it turns an attacker's unlimited guesses into a handful. Open the Local Security Policy (secpol.msc) or Group Policy and go to Account Policies > Account Lockout Policy. Set Account lockout threshold to 5 invalid attempts, Account lockout duration to 30 minutes, and Reset account lockout counter to 30 minutes. Now a bot gets five guesses per half hour per account — from thousands per minute to a trickle.

Two caveats. First, lockout can be weaponized for denial of service: an attacker who knows your username can deliberately lock you out. That is why this layer pairs with renaming the Administrator account and using non-obvious usernames — the attacker cannot lock what they cannot name. Second, on a shared or hosted server, confirm with your provider before setting aggressive lockouts; if you lock yourself out of the only admin account, you will need their help to get back in. Providers with 24/7 support can reset this for you, but prevention beats the ticket.

Layer 2: passwords attackers cannot guess

Lockout slows guessing; a strong password makes guessing pointless. The rule is simple: long, unique, and generated — not memorable. A 20-character random password is effectively immune to brute force within any realistic lockout regime. Use a password manager to generate and store server passwords; humans are terrible at both randomness and recall, and password reuse across servers turns one breach into many.

Equally important: rename the default Administrator account (Local Security Policy > Local Policies > Security Options > "Accounts: Rename administrator account") and disable the Guest account. Half of every brute force dictionary targets "Administrator" specifically — renaming it deletes half the attacker's playbook at a stroke. Create your own admin-privileged user with an unguessable name instead.

Defense layerWhat it stopsEffort
Account lockout (5 tries / 30 min)High-speed password guessing5 minutes in secpol.msc
Long unique passwordsDictionary and brute force successPassword manager setup
Rename Administrator, disable GuestUsername-targeted attacks2 minutes
IP allowlist in firewallEntire internet from reaching RDPOngoing IP management
Nonstandard RDP portDrive-by scans of port 3389See port-change guide
Network Level AuthenticationUnauthenticated session resource useCheckbox + client support
VPN-only RDP accessAll direct internet attacksVPN setup
No defenses thousands/hr + lockout ~10/hr + strong passwords ~0 success + IP allowlist or VPN: attackers cannot even reach the login Each layer multiplies the cost; the last layers remove the opportunity entirely.
Fig. 2 — How each layer cuts the attacker's effective guesses.

Layer 3: keep attackers from the login screen

The strongest defenses do not harden the login — they remove it from the internet. If you connect from a fixed home or office IP, add a Windows Firewall rule allowing port 3389 only from your IPs (see our Windows Firewall RDP rules guide). Overnight, your 4625 storm drops to zero, because the bots cannot reach the port at all. The maintenance cost is updating the rule when your IP changes — dynamic-DNS hostnames in firewall rules need a script, so many people allowlist a small set of static IPs instead.

Changing the RDP port away from 3389 cuts drive-by scans dramatically — most bots only knock on the default door. It is obscurity, not security, and a targeted attacker will port-scan you anyway, but as one layer among many it is cheap. The full walkthrough is in how to change the RDP port. Requiring NLA (our NLA guide) forces authentication before a session is even allocated, denying attackers the pre-auth attack surface. And the gold standard — VPN-only access, closing RDP to the internet entirely and connecting through a VPN — is covered in securing RDP with a VPN.

Harden your server: step by step

  1. Check Event Viewer for event ID 4625 to see the attack volume you are already receiving.
  2. Set the account lockout policy: secpol.msc > Account Lockout Policy > 5 attempts, 30-minute lockout and reset.
  3. Rename the Administrator account and disable Guest; create your own admin user with a non-obvious name.
  4. Set long unique passwords on every account with a password manager — 20+ random characters.
  5. Require Network Level Authentication on the server (System Properties > Remote > allow connections only from computers running Remote Desktop with NLA).
  6. Restrict RDP in Windows Firewall to your known IPs if your address is stable.
  7. Consider a nonstandard port as an additional layer — never as the only layer.
  8. Re-check the Security log weekly — attempts continue, but successes must stay at zero.

For the complete hardening picture beyond brute force, work through the RDP security best practices checklist and two-factor authentication for RDP. Every layer here assumes you have administrator rights on the server — one more reason to choose plans with full admin RDP access.

Need an RDP server that just works?

Full admin access, dedicated IP, 24/7 support and a 3-day money-back guarantee on every plan.

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

3-day money-back guarantee · Dedicated IP on every plan · 24/7 support

RDP Brute Force Protection: FAQ

Open Event Viewer, Windows Logs, Security, and filter for event ID 4625 (failed logons). Bursts of 4625s from foreign IPs against usernames like Administrator mean bots are actively guessing passwords against your server.

A widely used baseline is locking the account after 5 invalid attempts, with a 30-minute lockout duration and a 30-minute counter reset. It reduces an attacker from thousands of guesses per minute to a handful per half hour.

It stops opportunistic scans that only knock on port 3389, which is most background noise. It does not stop a targeted attacker who port-scans you, so treat it as one layer alongside lockout, strong passwords, and IP restrictions — never the only defense.

Yes. A huge share of brute force attempts target the username Administrator specifically. Renaming it (and disabling Guest) removes the attacker's most reliable username at almost zero cost.

Completely, for direct attacks: if RDP is closed to the internet and reachable only through the VPN, bots cannot reach the login screen at all. It is the strongest single layer, at the cost of setting up and maintaining the VPN.

Treat it as a compromise: disconnect the server from the network, change all passwords from a clean machine, review what the intruder touched, and consider rebuilding the server. Then apply every layer in this guide before exposing RDP again.