diff --git a/src/lib/editor/extensions/CLAUDE.md b/src/lib/editor/extensions/CLAUDE.md
index ea6bbf2..b1d7fb9 100644
--- a/src/lib/editor/extensions/CLAUDE.md
+++ b/src/lib/editor/extensions/CLAUDE.md
@@ -31,7 +31,7 @@ Five topics are large enough to have their own deep-dive — read the file befor
 - **`link.ts`** — hyperlink mark; owns `Mod-k`, which fires the event `Toolbar.svelte` listens for to open the link dialog. A `#name` href targets a bookmark: `Editor.svelte`'s `handleClick` scrolls to it instead of opening a tab. The `plain` attr (`data-plain`, `editor.css`) suppresses the editor's own blue: Word paints a link through its **Hyperlink character style**, which arrives as an ordinary mark, so every DOCX link is plain and one carrying no style is drawn like the text around it — while LibreOffice paints every ODF `text:a` styled or not (probed), so those never are. The DOCX exporter paints `#0563C1` only on a non-plain link, which is how a round trip tells its own links from a file's.
 - **`comment.ts`** — an annotation on a range of text, a mark like `bookmark`: the whole comment (author, date, body, resolved) rides its attrs, so two runs of one comment merge back together and no second store is needed. `commentRanges(doc)` walks every range (an id split by a paragraph boundary yields several, which is why the commands walk all of them), `comments(doc)` keeps one per id for `CommentsPane.svelte`. A point comment — LibreOffice's, made with nothing selected, written with no `office:name` and no end element — attaches on import to the run before it, else the one after: the editor's comment is a mark and needs a range.
 - **`trackChanges.ts`** — the `insertion`/`deletion` marks plus the recording plugin. `appendTransaction` marks whatever a transaction added, and `adjacentInsertion` reuses the id of the run just before it when the author matches, so a typed sentence is one change and not one per keystroke. A delete never reaches the document: `filterTransaction` rejects it and dispatches a mark pass instead (the author's own pending insertion is removed for real, as both word processors do). Every transaction of our own carries the `RECORDING` meta — accept/reject included, or its `tr.delete` would be read as a fresh user deletion and marked up again. `revisions(doc)` merges adjacent runs of one id, so an export writes one change per id.
-- **`caption.ts`** — a caption is an ordinary paragraph in the `Caption` style (LibreOffice's, probed: 10pt italic, 0.212cm above and below) holding one `sequenceField` atom, its running number. `sequenceFields(doc)` counts one counter per category in document order; the cached `number` attr is what an export writes, and an `appendTransaction` (off the undo stack) plus an `onCreate` pass keep it true — a file's own cached numbers are unreadable wherever the format is not arabic. The counter's name differs per format: ODF `Illustration`/`Table`, Word `Figure`/`Table`, and `ODF_SEQ_CATEGORY` reads both back. `CaptionDialog.svelte` (References tab) owns the insert.
+- **`caption.ts`** — a caption is an ordinary paragraph in the `Caption` style (LibreOffice's, probed: 10pt italic, 0.212cm above and below) holding one `sequenceField` atom, its running number. `sequenceFields(doc)` counts one counter per category in document order; the cached `number` attr is what an export writes, and an `appendTransaction` (off the undo stack) plus an `onCreate` pass keep it true — a file's own cached numbers are unreadable wherever the format is not arabic. The counter's name differs per format: ODF `Illustration`/`Table`, Word `Figure`/`Table`, and `ODF_SEQ_CATEGORY` reads both back. The caption paragraph takes the **alignment of the block it captions**, so it stands under a centred or right-aligned picture instead of at the margin — LibreOffice frames picture and caption together, and a loose paragraph can only carry that over as its alignment. `CaptionDialog.svelte` (References tab) owns the insert.
 - **`bookmark.ts`** / **`crossReference.ts`** — a named range of text (a mark) and the live references to it (`REF`/`PAGEREF`). `ToolbarExpanded.svelte` owns both dialogs, as it does the link one.
 - **`searchReplace.ts`** — Find & Replace: decorations highlight all matches, commands navigate/replace. Touches the document only for actual replacements. `useRegex` takes the term as a pattern instead of escaping it (a whole-word search then wraps it in a non-capturing group first, so the `\b` guards bound the alternation); each match keeps its captures so `expandGroups` can fill `$1`…`$9`/`$&` in the replacement.
 - **`spellCheck.ts`** — decorations from `spell/controller.ts`; `spellErrorAt` feeds the context menu.
diff --git a/src/lib/editor/extensions/caption.ts b/src/lib/editor/extensions/caption.ts
index 90e3da4..d9d1215 100644
--- a/src/lib/editor/extensions/caption.ts
+++ b/src/lib/editor/extensions/caption.ts
@@ -117,8 +117,11 @@ export const SequenceField = Node.create({
           content.push({ type: this.name, attrs: { category, format: '1', number: 1 } });
           const tail = `${separator}${text}`;
           if (tail) content.push({ type: 'text', text: tail });
+          // A caption stands under its picture, not at the margin: LibreOffice frames the
+          // two together, and the alignment is what carries that over to a loose paragraph.
+          const textAlign = $from.node(1).attrs.textAlign ?? null;
           return chain()
-            .insertContentAt(at, { type: 'paragraph', attrs: { styleName: CAPTION_STYLE }, content })
+            .insertContentAt(at, { type: 'paragraph', attrs: { styleName: CAPTION_STYLE, textAlign }, content })
             .focus()
             .run();
         },
diff --git a/tests/roundtrip.test.ts b/tests/roundtrip.test.ts
index b7b1794..9bce129 100644
--- a/tests/roundtrip.test.ts
+++ b/tests/roundtrip.test.ts
@@ -1958,10 +1958,11 @@ describe('Leg 18: per-paragraph text direction', () => {
 describe('Leg 19: captions (a sequence field per category)', () => {
   const SEQ = (category: string, number: number): N =>
     ({ type: 'sequenceField', attrs: { category, format: '1', number } });
-  const CAP = (category: string, label: string, n: number, text: string): N =>
-    P({ styleName: 'Caption' }, T(`${label} `), SEQ(category, n), T(`: ${text}`));
+  const CAP = (category: string, label: string, n: number, text: string, textAlign?: string): N =>
+    P({ styleName: 'Caption', ...(textAlign ? { textAlign } : {}) }, T(`${label} `), SEQ(category, n), T(`: ${text}`));
   const doc: N = { type: 'doc', content: [
-    CAP('figure', 'Figure', 1, 'first picture'),
+    // A caption under a centred picture is centred with it, which is what has to travel.
+    CAP('figure', 'Figure', 1, 'first picture', 'center'),
     CAP('table', 'Table', 1, 'first table'),
     CAP('figure', 'Figure', 2, 'second picture'),
   ] };
@@ -1980,6 +1981,7 @@ describe('Leg 19: captions (a sequence field per category)', () => {
     check('numbers round-trip', fields.map((f: N) => f?.attrs?.number).join() === '1,1,2',
       fields.map((f: N) => f?.attrs?.number));
     check('caption style kept', back.content[0].attrs?.styleName === 'Caption', back.content[0].attrs);
+    check('the picture’s alignment kept', back.content[0].attrs?.textAlign === 'center', back.content[0].attrs);
   });
 
   it('DOCX: a SEQ field per category, and back', async () => {
@@ -1993,6 +1995,7 @@ describe('Leg 19: captions (a sequence field per category)', () => {
       fields.map((f: N) => f?.attrs));
     check('numbers round-trip', fields.map((f: N) => f?.attrs?.number).join() === '1,1,2',
       fields.map((f: N) => f?.attrs?.number));
+    check('the picture’s alignment kept', back.content[0].attrs?.textAlign === 'center', back.content[0].attrs);
   });
 
   it('a document without captions declares no counter', async () => {
diff --git a/tests/unit/caption.test.ts b/tests/unit/caption.test.ts
index 24a64e7..0d3c37f 100644
--- a/tests/unit/caption.test.ts
+++ b/tests/unit/caption.test.ts
@@ -1,6 +1,11 @@
 import { describe, it, expect } from 'vitest';
 import { Schema } from '@tiptap/pm/model';
-import { sequenceFields, ODF_SEQ_CATEGORY, seqCategoryOf } from '../../src/lib/editor/extensions/caption';
+import { Editor } from '@tiptap/core';
+import Document from '@tiptap/extension-document';
+import Paragraph from '@tiptap/extension-paragraph';
+import Text from '@tiptap/extension-text';
+import TextAlign from '@tiptap/extension-text-align';
+import { sequenceFields, ODF_SEQ_CATEGORY, seqCategoryOf, SequenceField } from '../../src/lib/editor/extensions/caption';
 
 // The numbering rule both word processors apply: one counter per category, in document
 // order. A schema of its own keeps the check off the editor's full extension list.
@@ -42,6 +47,36 @@ describe('sequenceFields', () => {
   });
 });
 
+describe('insertCaption', () => {
+  const makeEditor = (textAlign: string | null) => {
+    const el = document.createElement('div');
+    document.body.appendChild(el);
+    return new Editor({
+      element: el,
+      extensions: [Document, Paragraph, Text, SequenceField, TextAlign.configure({ types: ['paragraph'] })],
+      content: { type: 'doc', content: [{ type: 'paragraph', attrs: { textAlign }, content: [{ type: 'text', text: 'picture' }] }] },
+    });
+  };
+  const insert = (editor: Editor) =>
+    editor.chain().focus().insertCaption({ category: 'figure', label: 'Figure', separator: ': ', text: 'x', above: false }).run();
+
+  it('takes the alignment of the block it captions, so it stands under the picture', () => {
+    const editor = makeEditor('center');
+    insert(editor);
+    const [, caption] = editor.getJSON().content as any[];
+    expect(caption.attrs.textAlign).toBe('center');
+    editor.destroy();
+  });
+
+  it('leaves the caption of an unaligned block unaligned', () => {
+    const editor = makeEditor(null);
+    insert(editor);
+    const [, caption] = editor.getJSON().content as any[];
+    expect(caption.attrs.textAlign ?? null).toBe(null);
+    editor.destroy();
+  });
+});
+
 describe('category names', () => {
   it('reads both spellings of the picture counter', () => {
     expect(ODF_SEQ_CATEGORY.Illustration).toBe('figure'); // LibreOffice's