diff --git a/packages/lexical-markdown/src/MarkdownImport.ts b/packages/lexical-markdown/src/MarkdownImport.ts
index e5ced6d006b..29c15742f51 100644
--- a/packages/lexical-markdown/src/MarkdownImport.ts
+++ b/packages/lexical-markdown/src/MarkdownImport.ts
@@ -292,7 +292,8 @@ function $importBlocks(
 function $normalizeMarkdownTextNode(textNode: TextNode): void {
   // A TabNode is a TextNode whose content is a tab, so without this guard the
   // rebuild below would destroy it and create an equivalent one in its place.
-  if ($isTabNode(textNode)) {
+  // Inline code keeps its tabs as text, so they stay inside the code span.
+  if ($isTabNode(textNode) || textNode.hasFormat('code')) {
     return;
   }
   const text = textNode.getTextContent();
diff --git a/packages/lexical-markdown/src/__tests__/unit/CodeSpanWhitespaceImport.test.ts b/packages/lexical-markdown/src/__tests__/unit/CodeSpanWhitespaceImport.test.ts
new file mode 100644
index 00000000000..964cf5b1616
--- /dev/null
+++ b/packages/lexical-markdown/src/__tests__/unit/CodeSpanWhitespaceImport.test.ts
@@ -0,0 +1,106 @@
+/**
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
+ *
+ * This source code is licensed under the MIT license found in the
+ * LICENSE file in the root directory of this source tree.
+ *
+ */
+
+// @vitest-environment node
+
+import {buildEditorFromExtensions} from '@lexical/extension';
+import {
+  $convertFromMarkdownString,
+  $convertToMarkdownString,
+  TRANSFORMERS,
+} from '@lexical/markdown';
+import {$getRoot, $isElementNode, $isTabNode, $isTextNode} from 'lexical';
+import {describe, expect, it} from 'vitest';
+
+import {MarkdownTestExtension} from '../utils';
+
+type Inline = {code: boolean; tab: boolean; text: string};
+
+function importParagraphs(markdown: string): Inline[][] {
+  using editor = buildEditorFromExtensions([MarkdownTestExtension]);
+  let paragraphs: Inline[][] = [];
+  editor.update(
+    () => {
+      $convertFromMarkdownString(markdown, TRANSFORMERS);
+      paragraphs = $getRoot()
+        .getChildren()
+        .map(block =>
+          $isElementNode(block)
+            ? block.getChildren().map(child => ({
+                code: $isTextNode(child) && child.hasFormat('code'),
+                tab: $isTabNode(child),
+                text: child.getTextContent(),
+              }))
+            : [],
+        );
+    },
+    {discrete: true},
+  );
+  return paragraphs;
+}
+
+function roundTrip(markdown: string): string {
+  using editor = buildEditorFromExtensions([MarkdownTestExtension]);
+  editor.update(
+    () => {
+      $convertFromMarkdownString(markdown, TRANSFORMERS);
+    },
+    {discrete: true},
+  );
+  return editor.read(() => $convertToMarkdownString(TRANSFORMERS));
+}
+
+const code = (text: string): Inline => ({code: true, tab: false, text});
+const text = (value: string): Inline => ({
+  code: false,
+  tab: false,
+  text: value,
+});
+const tab: Inline = {code: false, tab: true, text: '\t'};
+
+describe('whitespace inside inline code (#9409)', () => {
+  it('keeps a tab inside a code span as part of the code', () => {
+    expect(importParagraphs('`a\tb`')).toEqual([[code('a\tb')]]);
+    expect(importParagraphs('a `\t` b')).toEqual([
+      [text('a '), code('\t'), text(' b')],
+    ]);
+    expect(importParagraphs('`\t`')).toEqual([[code('\t')]]);
+  });
+
+  it('round trips a code span that holds a tab', () => {
+    expect(roundTrip('`a\tb`')).toBe('`a\tb`');
+    // A span that starts and ends with whitespace is padded, and a reader
+    // strips that padding back to the tab.
+    expect(roundTrip('`\t`')).toBe('` \t `');
+    expect(roundTrip('` \t `')).toBe('` \t `');
+    expect(roundTrip('a `\t` b')).toBe('a ` \t ` b');
+    expect(roundTrip('a ` \t ` b')).toBe('a ` \t ` b');
+  });
+
+  it('keeps a paragraph whose only content is whitespace code', () => {
+    expect(importParagraphs('x\n\n` `\n\ny')).toEqual([
+      [text('x')],
+      [code(' ')],
+      [text('y')],
+    ]);
+    expect(importParagraphs('x\n\n`\t`\n\ny')).toEqual([
+      [text('x')],
+      [code('\t')],
+      [text('y')],
+    ]);
+  });
+
+  it('still turns tabs outside code into tab nodes', () => {
+    expect(importParagraphs('a\tb')).toEqual([[text('a'), tab, text('b')]]);
+    expect(importParagraphs('a\t`b`')).toEqual([[text('a'), tab, code('b')]]);
+  });
+
+  it('still drops empty paragraphs', () => {
+    expect(importParagraphs('x\n\n\n\ny')).toEqual([[text('x')], [text('y')]]);
+  });
+});
diff --git a/packages/lexical-markdown/src/utils.ts b/packages/lexical-markdown/src/utils.ts
index 200b5c10b56..893f6f8fc29 100644
--- a/packages/lexical-markdown/src/utils.ts
+++ b/packages/lexical-markdown/src/utils.ts
@@ -72,6 +72,8 @@ export function isEmptyParagraph(node: LexicalNode): boolean {
     firstChild == null ||
     (node.getChildrenSize() === 1 &&
       $isTextNode(firstChild) &&
+      // Inline code is content even when it holds only whitespace.
+      !firstChild.hasFormat('code') &&
       MARKDOWN_EMPTY_LINE_REG_EXP.test(firstChild.getTextContent()))
   );
 }