diff --git a/.github/styles/base/Dictionary.txt b/.github/styles/base/Dictionary.txt
index 8d1266def5..eb98af7445 100644
--- a/.github/styles/base/Dictionary.txt
+++ b/.github/styles/base/Dictionary.txt
@@ -158,6 +158,7 @@ cors
 cosmiconfig
 cosocket
 CPUs
+vCPUs
 crd
 crds
 created_at
@@ -677,6 +678,7 @@ perf
 performant
 petstore
 PDFs
+pg_max_concurrent_queries
 pgdump
 pgvector
 pipeline
diff --git a/app/_support/pg_max_concurrent_queries-per-worker-process.md b/app/_support/pg_max_concurrent_queries-per-worker-process.md
new file mode 100644
index 0000000000..c0fecff750
--- /dev/null
+++ b/app/_support/pg_max_concurrent_queries-per-worker-process.md
@@ -0,0 +1,110 @@
+---
+title: "pg_max_concurrent_queries is per worker process, not per node"
+content_type: support
+description: "How to limit the total number of PostgreSQL database connections when running {{site.base_gateway}} on multiple nodes in Kubernetes."
+products:
+  - gateway
+
+works_on:
+  - on-prem
+  - konnect
+
+related_resources:
+  - text: "{{site.base_gateway}} configuration reference - `pg_max_concurrent_queries`"
+    url: /gateway/configuration/#pg-max-concurrent-queries
+  - text: "{{site.base_gateway}} configuration reference - `nginx_worker_processes`"
+    url: /gateway/configuration/#nginx-worker-processes
+
+tldr:
+  q: Why does my PostgreSQL connection count exceed the limit I set with pg_max_concurrent_queries?
+  a: |
+    `pg_max_concurrent_queries` is applied **per nginx worker process**, so the actual maximum
+    number of concurrent database connections is:
+
+    ```
+    pg_max_concurrent_queries × nginx_worker_processes × number of Kong nodes
+    ```
+
+    To bring the total under your database connection limit, explicitly set
+    `nginx_worker_processes` to a fixed integer value rather than relying on the
+    default `auto` setting, which bases the count on the host's vCPUs and can
+    produce more workers than expected in Kubernetes.
+
+---
+
+## Problem
+
+The PostgreSQL database used by {{site.base_gateway}} has a maximum number of permitted
+connections. After setting `pg_max_concurrent_queries` to a value below that maximum,
+the connection limit is still exceeded. {{site.base_gateway}} is running on multiple
+nodes deployed in Kubernetes.
+
+## Cause
+
+The `pg_max_concurrent_queries` setting is scoped **per nginx worker process**, not
+per Kong node. The actual maximum number of concurrent database connections is therefore:
+
+```
+pg_max_concurrent_queries × nginx_worker_processes × number of Kong nodes
+```
+
+The `nginx_worker_processes` parameter defaults to `auto`, which sets the number of
+worker processes equal to the number of available vCPUs. In a Kubernetes environment,
+the calculation is based on the **host** vCPUs (not the container's CPU limit), which
+can result in significantly more workers — and therefore more database connections —
+than expected.
+
+## Solution
+
+1. Determine the current number of nginx worker processes on a running Kong node:
+
+   ```bash
+   ps aux | grep "[n]ginx: worker process" | wc -l
+   ```
+
+2. Calculate the maximum connection count you need to stay within your database limit.
+   For example, if your database allows 200 connections, you have 4 Kong nodes, and
+   you want `pg_max_concurrent_queries` set to 10:
+
+   ```
+   200 connections ÷ (4 nodes × 10 queries) = 5 worker processes per node
+   ```
+
+3. Explicitly set `nginx_worker_processes` to a fixed integer in `kong.conf`:
+
+   ```
+   nginx_worker_processes = 5
+   ```
+
+   Or via an environment variable:
+
+   ```bash
+   KONG_NGINX_WORKER_PROCESSES=5
+   ```
+
+4. Reload or restart {{site.base_gateway}} for the change to take effect:
+
+   ```bash
+   kong reload
+   ```
+
+5. Verify the worker count reflects the new value:
+
+   ```bash
+   ps aux | grep "[n]ginx: worker process" | wc -l
+   ```
+
+## Validation
+
+After restarting, confirm the total connection count stays within bounds by monitoring
+your database connection metrics. For PostgreSQL, you can check active connections with:
+
+```sql
+SELECT count(*) FROM pg_stat_activity WHERE datname = '<your_kong_database>';
+```
+
+The count should remain at or below:
+
+```
+pg_max_concurrent_queries × nginx_worker_processes × number of Kong nodes
+```