diff --git a/docs/assets/read_batch.png b/docs/assets/read_batch.png
new file mode 100644
index 0000000000..47c2de115f
Binary files /dev/null and b/docs/assets/read_batch.png differ
diff --git a/docs/user-guide/reference/configuration/read-batch-size.md b/docs/user-guide/reference/configuration/read-batch-size.md
new file mode 100644
index 0000000000..de3a93fdd6
--- /dev/null
+++ b/docs/user-guide/reference/configuration/read-batch-size.md
@@ -0,0 +1,25 @@
+### Understanding `readBatchSize`
+
+At its core, `readBatchSize` is the **maximum number of messages a single vertex replica (pod) will try to read per cycle**. 
+
+
+#### **Default Behavior (Single Partition)**
+
+By default, the edge connecting vertices has one partition. In this common setup, a single replica reads from that single partition.
+
+**Note**: Monovertex always follows this default behavior, as it does not have the concept of partitions.
+* **Example:** If `readBatchSize` is **500** and you have **1** replica, that replica will read up to **500** messages in each processing cycle. Simple!
+
+#### **Behavior with Multi-Partitioned Edges**
+
+When you configure a vertex to have multiple partitions (**not applicable for Source Vertices**), you increase its ability to process data in parallel. The key thing to understand is that the `readBatchSize` limit applies to the read operation on **each partition** that a replica is responsible for.
+
+This means the total number of messages a single replica can process per cycle becomes:
+$$\text{Total Messages per Replica} = \text{readBatchSize} \times \text{Number of Partitions}$$
+
+* **Example:** Let's say you configure a vertex to have **3** partitions and it's running with **1** replica. If `readBatchSize` is **500**:
+    * The single replica will read up to **500** messages from Partition 1.
+    * It will *also* read up to **500** messages from Partition 2 and 3.
+    * In total, that single replica can now process up to **1,500** messages per cycle (500 * 3).
+
+![read_batch.png](../../../assets/read_batch.png)
diff --git a/docs/user-guide/reference/pipeline-tuning.md b/docs/user-guide/reference/pipeline-tuning.md
index 8dcaaa2783..02751d1a44 100644
--- a/docs/user-guide/reference/pipeline-tuning.md
+++ b/docs/user-guide/reference/pipeline-tuning.md
@@ -5,15 +5,25 @@ In a data processing [pipeline](../../core-concepts/pipeline.md), certain parame
 ## Vertex Tuning
 
 Each [vertex](../../core-concepts/vertex.md) keeps running the cycle of reading data from an Inter-Step Buffer (or data source),
-processing the data, and writing to the next Inter-Step Buffers (or sinks). There are some parameters can be adjusted for this data
+processing the data, and writing to the next Inter-Step Buffers (or sinks). There are some parameters that can be adjusted for this data
 processing cycle.
 
-- `readBatchSize` - The number of messages to read in each cycle, with a default value of `500`. It works together with `readTimeout` during a read operation, concluding when either limit is reached first.
-- `readTimeout` - Read timeout from the source or Inter-Step Buffer, defaults to `1s`. It works in conjunction with `readBatchSize`.
-- `bufferMaxLength` - How many unprocessed messages can be existing in the Inter-Step Buffer, defaults to `30000`.
-- `bufferUsageLimit` - The percentage of the buffer usage limit, a valid number should be less than 100. Default value is `80`, which means `80%`.
+- **`readBatchSize`** — Maximum number of messages **each vertex replica** reads **per cycle**.
+  *Default:* `500`. A cycle ends when either this cap is reached **or** `readTimeout` expires.
+  
+**NOTE**: If you are working with [Multi-Partitioned edges](multi-partition.md), please refer to the advanced
+  [configuration](./configuration/read-batch-size.md) for `readBatchSize`.
 
-These parameters can be customized under `spec.limits` as below, once defined, they apply to all the vertices and Inter-Step Buffers of the pipeline.
+- **`readTimeout`** — Read timeout from the source or Inter-Step Buffer.
+  *Default:* `1s`. Works in conjunction with `readBatchSize` (whichever condition is met first ends the cycle).
+
+- **`bufferMaxLength`** — How many unprocessed messages can exist in the Inter-Step Buffer.
+  *Default:* `30000`.
+
+- **`bufferUsageLimit`** — The percentage threshold for buffer usage; must be `< 100`.
+  *Default:* `80` (i.e., `80%`).
+
+These parameters can be customized under `spec.limits` as below; once defined, they apply to all the vertices and Inter-Step Buffers of the pipeline.
 
 ```yaml
 apiVersion: numaflow.numaproj.io/v1alpha1
diff --git a/mkdocs.yml b/mkdocs.yml
index 3e3f86726f..56c799f55d 100644
--- a/mkdocs.yml
+++ b/mkdocs.yml
@@ -119,6 +119,7 @@ nav:
               - user-guide/reference/configuration/pipeline-customization.md
               - user-guide/reference/configuration/dra.md
               - user-guide/reference/configuration/istio.md
+              - user-guide/reference/configuration/read-batch-size.md
               - user-guide/reference/configuration/max-message-size.md
               - user-guide/reference/configuration/update-strategy.md
           - user-guide/reference/kustomize/kustomize.md