diff --git a/tildes/tests/test_markdown.py b/tildes/tests/test_markdown.py
index 63c0995..333110f 100644
--- a/tildes/tests/test_markdown.py
+++ b/tildes/tests/test_markdown.py
@@ -81,6 +81,22 @@ def test_accidental_ordered_list():
     assert "<ol" not in html
 
 
+def test_merged_blockquotes():
+    """Ensure subsequent blockquotes are merged."""
+    markdown = "> Paragraph 1\n\n> Paragraph 2"
+    html = convert_markdown_to_safe_html(markdown)
+
+    assert html.count("<blockquote") == 1
+
+
+def test_separated_blockquotes():
+    """Ensure blockquotes can be separated with an additional newline."""
+    markdown = "> Paragraph 1\n\n\n> Paragraph 2"
+    html = convert_markdown_to_safe_html(markdown)
+
+    assert html.count("<blockquote") == 2
+
+
 def test_existing_newline_not_doubled():
     """Ensure that the standard markdown line break doesn't result in two."""
     markdown = "A deliberate line  \nbreak"
diff --git a/tildes/tildes/lib/markdown.py b/tildes/tildes/lib/markdown.py
index 5b2cca8..cf2f125 100644
--- a/tildes/tildes/lib/markdown.py
+++ b/tildes/tildes/lib/markdown.py
@@ -107,6 +107,8 @@ BAD_ORDERED_LIST_REGEX = re.compile(
 
 STRIP_IMAGE_ELEMENTS_REGEX = re.compile(r'<img src="([^"<>]*?)" alt="([^"<>]*?)" />')
 
+SUBSEQUENT_BLOCKQUOTES_REGEX = re.compile("^>([^\n]*?)\n\n(?=>)", flags=re.MULTILINE)
+
 
 @histogram_timer("markdown_processing")
 def convert_markdown_to_safe_html(markdown: str) -> str:
@@ -143,12 +145,26 @@ def preprocess_markdown(markdown: str) -> str:
     """Pre-process markdown before passing it to CommonMark."""
     markdown = escape_accidental_ordered_lists(markdown)
 
+    markdown = merge_subsequent_blockquotes(markdown)
+
     # fix the "shrug" emoji ¯\_(ツ)_/¯ to prevent markdown mangling it
     markdown = markdown.replace(r"¯\_(ツ)_/¯", r"¯\\\_(ツ)\_/¯")
 
     return markdown
 
 
+def merge_subsequent_blockquotes(markdown: str) -> str:
+    """Merge subsequent (separated) blockquotes into a single one.
+
+    By default, if someone quotes more than one paragraph without also adding the >
+    symbol on the blank lines between them, they will be interpreted as separate
+    blockquotes. This almost never seems to be intended, so this merges them. If
+    separate quotes are wanted, they can still achieve it by using at least two
+    newlines between quoted paragraphs (or various other methods).
+    """
+    return SUBSEQUENT_BLOCKQUOTES_REGEX.sub(">\\1\n>\n", markdown)
+
+
 def escape_accidental_ordered_lists(markdown: str) -> str:
     """Escape markdown that's probably an accidental ordered list.