RDP session timeout settings control what Windows does with idle, disconnected, and long-running remote sessions — and the defaults surprise a lot of admins. The direct answer: open Group Policy → Computer Configuration → Administrative Templates → Windows Components → Remote Desktop Services → Remote Desktop Session Host → Session Time Limits, where you can set separate limits for active, idle, and disconnected sessions, plus what happens when any limit is hit. This guide explains each timer, the recommended values for different environments, and how to stop the classic "my session got logged off overnight" complaint.

The three timers (they're independent)

Windows tracks three different clocks per session, and each has its own policy:

  • Active session limit — maximum total connected time, regardless of activity. When it expires, the session is disconnected or logged off (your choice).
  • Idle session limit — how long a connected session may sit with no keyboard/mouse input. This is the one that reclaims resources from people who go home without logging off.
  • Disconnected session limit — how long a session survives after the user closes the RDP client without logging off. The programs keep running during this window, which is why closing the window "to keep things running" works — until this timer fires.

Confusing these is the root of most timeout tickets. A user who complains "I got kicked out" usually hit the idle limit; a user whose overnight job died hit the disconnected limit after closing the laptop lid.

Logon Goes idle idle timer starts Disconnects disconnected timer starts Limit hit log off / disconnect Active session limit spans the whole timeline
Fig. 1 — The three timers run on different triggers. Know which one fired before you change anything.

Configuring the limits step by step

  1. On the server, press Win + R, type gpedit.msc, and press Enter (on domain environments, use Group Policy Management on the domain controller instead).
  2. Navigate to Computer Configuration → Administrative Templates → Windows Components → Remote Desktop Services → Remote Desktop Session Host → Session Time Limits.
  3. Open "Set time limit for active but idle Remote Desktop Services sessions" → Enabled → set your idle limit (e.g., 2 hours) → OK.
  4. Open "Set time limit for disconnected sessions" → Enabled → set the disconnected limit (e.g., 8 hours for overnight jobs, 1 hour for shared terminals) → OK.
  5. Optionally set "Set time limit for active Remote Desktop Services sessions" if you want a hard cap on total session length (common on kiosk/shared machines, rare on personal servers).
  6. Open "End session when time limits are reached": Enabled means the session is logged off (programs closed); Disabled/Not Configured means it's merely disconnected (programs keep running, session can be reconnected).
  7. Run gpupdate /force in an elevated Command Prompt, then verify with gpresult /r or rsop.msc.

The single most important choice is step 6. Disconnect ≠ log off. Disconnecting leaves everything running and is what most people actually want; logging off kills unsaved work. Set "End session when time limits are reached" to Disabled unless you have a specific reason to force logoffs.

EnvironmentIdle limitDisconnected limitOn limit reached
Personal RDP serverNever (or 24 h)NeverDisconnect
Small team server4 hours8–12 hoursDisconnect
Shared terminal / kiosk30–60 minutes15–30 minutesLog off
Overnight batch jobsNeverNever (or > job runtime)Disconnect
Regulated / compliance15 minutes15 minutesLog off

These are starting examples — tune to your own policy. The pattern to notice: disconnect is the gentle default everywhere except shared and regulated machines, where forced logoff protects the next user and the audit trail.

Disconnect programs keep running reconnect resumes all gentle default ✓ Log off programs closed unsaved work lost use deliberately
Fig. 2 — Disconnect preserves the session; logoff ends it. Choose per environment, not per habit.

The overnight-job pattern (done right)

The classic use case: you start a long render, backtest, or download in the RDP session, close the client, and expect it running in the morning. For this to work reliably you need three things aligned: the disconnected session limit set to Never (or longer than the job), "End session when time limits are reached" not forcing logoff, and the server's power settings set to never sleep (a sleeping server pauses everything regardless of session policy).

Also confirm the job itself survives a disconnect — nearly all do, since disconnect only detaches the display. The exception is software that needs an active GUI session (some screen-capture or GUI-automation tools); test those once during the day before trusting them overnight. If jobs die specifically when the client closes, check the disconnected limit first — it's the culprit far more often than the application.

On a rented server you control all of this: every RDP Temple plan (Shared $6/month, Virtual $28/month, Dedicated $165/month) includes full admin RDP, so gpedit.msc is yours to configure, with a dedicated IP and 24/7 support if a policy change ever locks you out. Pair sensible timeouts with the tuning in our performance monitoring guide and session settings stay a set-and-forget affair — and if your sessions drop for network reasons rather than policy, that's covered under reducing RDP latency.

When policies collide: precedence and verification

Session timeout settings can come from local Group Policy, domain Group Policy, the RDS deployment properties, and even per-user settings in Active Directory — and when they disagree, the winner isn't always obvious. The precedence order that matters:

  • Domain GPO beats local GPO. If the machine is domain-joined, a domain policy on Session Time Limits overrides whatever you set in gpedit.msc locally. Your local change will appear to "not stick" — it was overridden, not ignored.
  • Computer Configuration beats User Configuration for these particular settings; session time limits live under Computer Configuration for a reason.
  • RDS collection settings (set in Server Manager → Remote Desktop Services → Collections → Tasks → Edit Properties → Session) can also impose limits, and on a full RDS deployment these are the settings most admins actually manage.

When behavior doesn't match what you configured, verify instead of guessing: run gpresult /h report.html from an elevated prompt and open the report — it shows exactly which GPO won each setting. For a quick check, rsop.msc shows the resultant policy live. And remember that policy changes need gpupdate /force plus, for some settings, a fresh session before they take effect — testing with the same already-connected session is a classic false negative.

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 Session Timeout Settings: FAQ

Set the disconnected session limit to Never in Group Policy under Remote Desktop Session Host then Session Time Limits, and make sure End session when time limits are reached is not forcing a logoff.

Disconnecting leaves your programs running so you can reconnect and resume. Logging off closes everything and any unsaved work is lost.

In Group Policy: Computer Configuration, Administrative Templates, Windows Components, Remote Desktop Services, Remote Desktop Session Host, Session Time Limits. Apply changes with gpupdate /force.

The disconnected session limit fired after you closed the client, or the server went to sleep. Set the disconnected limit longer than the job and disable sleep in the power settings.

As starting examples: never on a personal server, a few hours on a team server, 30 to 60 minutes on shared terminals, and 15 minutes in regulated environments.

Yes. Session timers are enforced server-side, so they apply regardless of which client connected. Browser tabs get abandoned often, so timeouts matter even more for web client users.