diff --git a/src/parquet-writer.js b/src/parquet-writer.js
index 41f5601..4c6ff20 100644
--- a/src/parquet-writer.js
+++ b/src/parquet-writer.js
@@ -34,13 +34,12 @@ export function ParquetWriter({ writer, schema, compressed = true, statistics =
  *
  * @param {object} options
  * @param {ColumnSource[]} options.columnData
- * @param {number} [options.rowGroupSize]
+ * @param {number | number[]} [options.rowGroupSize]
  */
 ParquetWriter.prototype.write = function({ columnData, rowGroupSize = 100000 }) {
   const columnDataRows = columnData[0]?.data?.length || 0
-  for (let groupStartIndex = 0; groupStartIndex < columnDataRows; groupStartIndex += rowGroupSize) {
+  for (const { groupStartIndex, groupSize } of groupIterator({ columnDataRows, rowGroupSize })) {
     const groupStartOffset = this.writer.offset
-    const groupSize = Math.min(rowGroupSize, columnDataRows - groupStartIndex)
 
     // row group columns
     /** @type {ColumnChunk[]} */
@@ -93,3 +92,32 @@ ParquetWriter.prototype.finish = function() {
   this.writer.appendUint32(0x31524150)
   this.writer.finish()
 }
+
+/**
+ * Create an iterator for row groups based on the specified row group size.
+ * If rowGroupSize is an array, it will return groups based on the sizes in the array.
+ * When the array runs out, it will continue with the last size.
+ *
+ * @param {object} options
+ * @param {number} options.columnDataRows - Total number of rows in the column data
+ * @param {number | number[]} options.rowGroupSize - Size of each row group or an array of sizes
+ * @returns {Array<{groupStartIndex: number, groupSize: number}>}
+ */
+function groupIterator({ columnDataRows, rowGroupSize }) {
+  if (Array.isArray(rowGroupSize) && !rowGroupSize.length) {
+    throw new Error('rowGroupSize array cannot be empty')
+  }
+  const groups = []
+  let groupIndex = 0
+  let groupStartIndex = 0
+  while (groupStartIndex < columnDataRows) {
+    const size = Array.isArray(rowGroupSize)
+      ? rowGroupSize[Math.min(groupIndex, rowGroupSize.length - 1)]
+      : rowGroupSize
+    const groupSize = Math.min(size, columnDataRows - groupStartIndex)
+    groups.push({ groupStartIndex, groupSize })
+    groupStartIndex += size
+    groupIndex++
+  }
+  return groups
+}
diff --git a/src/types.d.ts b/src/types.d.ts
index b01b870..f5a168e 100644
--- a/src/types.d.ts
+++ b/src/types.d.ts
@@ -20,7 +20,7 @@ export interface ParquetWriteOptions {
   schema?: SchemaElement[]
   compressed?: boolean
   statistics?: boolean
-  rowGroupSize?: number
+  rowGroupSize?: number | number[]
   kvMetadata?: KeyValue[]
 }
 
diff --git a/test/write.buffer.test.js b/test/write.buffer.test.js
index 4b875b6..0099307 100644
--- a/test/write.buffer.test.js
+++ b/test/write.buffer.test.js
@@ -264,6 +264,26 @@ describe('parquetWriteBuffer', () => {
     expect(result[199]).toEqual({ int: 13 })
   })
 
+  it('splits row groups with custom sizes', async () => {
+    const data = Array(200).fill(13)
+    const file = parquetWriteBuffer({ columnData: [{ name: 'int', data }], rowGroupSize: [20, 50] })
+    const metadata = parquetMetadata(file)
+    expect(metadata.row_groups.length).toBe(5)
+    expect(metadata.row_groups[0].num_rows).toBe(20n)
+    expect(metadata.row_groups[1].num_rows).toBe(50n)
+    // should use last size for remaining row groups
+    expect(metadata.row_groups[2].num_rows).toBe(50n)
+    expect(metadata.row_groups[3].num_rows).toBe(50n)
+    expect(metadata.row_groups[4].num_rows).toBe(30n)
+    // round trip
+    const result = await parquetReadObjects({ file })
+    expect(result.length).toBe(200)
+    expect(result[0]).toEqual({ int: 13 })
+    expect(result[49]).toEqual({ int: 13 })
+    expect(result[50]).toEqual({ int: 13 })
+    expect(result[199]).toEqual({ int: 13 })
+  })
+
   it('throws for wrong type specified', () => {
     expect(() => parquetWriteBuffer({ columnData: [{ name: 'int', data: [1, 2, 3], type: 'INT64' }] }))
       .toThrow('parquet expected bigint value')