diff --git a/README.md b/README.md
index cf344c9..b7e49bc 100644
--- a/README.md
+++ b/README.md
@@ -117,6 +117,8 @@ Schema migrations run automatically on first boot — no manual migration comman
 | `SSL_KEY` | Path to SSL private key | `server/certs/key.pem` |
 | `PING_INTERVAL` | Socket.IO Engine.IO ping interval (ms). Raise for slow TV WebKits that miss pongs under decode load. | `30000` |
 | `PING_TIMEOUT` | Socket.IO Engine.IO pong wait (ms). Lower = faster dead-socket detection; higher = more forgiving of laggy clients. | `30000` |
+| `HEARTBEAT_INTERVAL` | App-level offline-checker frequency (ms). How often the server sweeps the device list looking for stale heartbeats. | `10000` |
+| `HEARTBEAT_TIMEOUT` | How long without an app-level heartbeat (ms) before marking a device offline. Raise for slow/jittery networks. | `45000` |
 
 ### Optional Integrations
 
diff --git a/server/config.js b/server/config.js
index 946e72a..405698b 100644
--- a/server/config.js
+++ b/server/config.js
@@ -8,8 +8,12 @@ module.exports = {
   contentDir: path.join(__dirname, 'uploads', 'content'),
   screenshotsDir: path.join(__dirname, 'uploads', 'screenshots'),
   frontendDir: path.join(__dirname, '..', 'frontend'),
-  heartbeatInterval: 10000,    // Check every 10s
-  heartbeatTimeout: 45000,     // Offline after 45s (3 missed 15s beats)
+  // App-level heartbeat. Checker runs every heartbeatInterval and marks
+  // devices offline if last_heartbeat is older than heartbeatTimeout.
+  // Env override for self-hosters on slow/jittery networks (issue #3:
+  // reporter found raising HEARTBEAT_TIMEOUT to 60s reduced false offlines).
+  heartbeatInterval: parseInt(process.env.HEARTBEAT_INTERVAL) || 10000,
+  heartbeatTimeout:  parseInt(process.env.HEARTBEAT_TIMEOUT)  || 45000,
   // Engine.IO transport-level ping/pong. Raised from Socket.IO defaults
   // (25000/20000) because TV WebKits (LG webOS, older Tizen) miss pongs
   // under decode load - tighter values cause spurious transport drops.