diff --git a/.changeset/afraid-lobsters-complain.md b/.changeset/afraid-lobsters-complain.md
new file mode 100644
index 0000000..6afb594
--- /dev/null
+++ b/.changeset/afraid-lobsters-complain.md
@@ -0,0 +1,5 @@
+---
+'markdown-to-jsx': patch
+---
+
+Fix false detection of tables in some scenarios.
diff --git a/index.compiler.spec.tsx b/index.compiler.spec.tsx
index d7296c8..0292923 100644
--- a/index.compiler.spec.tsx
+++ b/index.compiler.spec.tsx
@@ -756,6 +756,16 @@ describe('headings', () => {
       </h1>
     `)
   })
+
+  it('#595 regression - handle pipe character inside header', () => {
+    render(compiler('# Heading | text'))
+
+    expect(root.innerHTML).toMatchInlineSnapshot(`
+      <h1 id="heading--text">
+        Heading | text
+      </h1>
+    `)
+  })
 })
 
 describe('images', () => {
@@ -1851,7 +1861,7 @@ describe('GFM tables', () => {
   it('should handle a basic table', () => {
     render(
       compiler(theredoc`
-        foo|bar
+        |foo|bar|
         ---|---
         1  |2
       `)
@@ -1886,7 +1896,7 @@ describe('GFM tables', () => {
   it('should handle a table with aligned columns', () => {
     render(
       compiler(theredoc`
-        foo|bar|baz
+        |foo|bar|baz|
         --:|:---:|:--
         1|2|3
       `)
diff --git a/index.tsx b/index.tsx
index b238a63..445040e 100644
--- a/index.tsx
+++ b/index.tsx
@@ -273,8 +273,7 @@ const LINK_AUTOLINK_BARE_URL_R = /^(https?:\/\/[^\s<]+[^<.,:;"')\]\s])/
 const LINK_AUTOLINK_MAILTO_R = /^<([^ >]+@[^ >]+)>/
 const LINK_AUTOLINK_R = /^<([^ >]+:\/[^ >]+)>/
 const CAPTURE_LETTER_AFTER_HYPHEN = /-([a-z])?/gi
-const NP_TABLE_R =
-  /^(.*\|.*)\n(?: *(\|? *[-:]+ *\|[-| :]*)\n((?:.*\|.*\n)*))?\n?/
+const NP_TABLE_R = /^(\|.*)\n(?: *(\|? *[-:]+ *\|[-| :]*)\n((?:.*\|.*\n)*))?\n?/
 const PARAGRAPH_R = /^[^\n]+(?:  \n|\n{2,})/
 const REFERENCE_IMAGE_OR_LINK = /^\[([^\]]*)\]:\s+<?([^\s>]+)>?\s*("([^"]*)")?/
 const REFERENCE_IMAGE_R = /^!\[([^\]]*)\] ?\[([^\]]*)\]/
@@ -926,20 +925,11 @@ function anyScopeRegex(regex: RegExp) {
   }
 }
 
-function matchParagraph(
-  source: string,
-  state: MarkdownToJSX.State,
-  prevCapturedString?: string
-) {
+function matchParagraph(source: string, state: MarkdownToJSX.State) {
   if (state.inline || state.simple) {
     return null
   }
 
-  if (prevCapturedString && !prevCapturedString.endsWith('\n')) {
-    // don't match continuation of a line
-    return null
-  }
-
   let match = ''
 
   source.split('\n').every(line => {