diff --git a/Lib/cgi.py b/Lib/cgi.py
--- a/Lib/cgi.py
+++ b/Lib/cgi.py
@@ -543,17 +543,17 @@ class FieldStorage:
         if 'content-length' in self.headers:
             try:
                 clen = int(self.headers['content-length'])
             except ValueError:
                 pass
             if maxlen and clen > maxlen:
                 raise ValueError('Maximum content length exceeded')
         self.length = clen
-        if self.limit is None and clen:
+        if self.limit is None and clen >= 0:
             self.limit = clen
 
         self.list = self.file = None
         self.done = 0
         if ctype == 'application/x-www-form-urlencoded':
             self.read_urlencoded()
         elif ctype[:10] == 'multipart/':
             self.read_multi(environ, keep_blank_values, strict_parsing)
@@ -720,19 +720,20 @@ class FieldStorage:
             self.bytes_read += len(hdr_text)
             parser.feed(hdr_text.decode(self.encoding, self.errors))
             headers = parser.close()
 
             # Some clients add Content-Length for part headers, ignore them
             if 'content-length' in headers:
                 del headers['content-length']
 
+            limit = (self.limit - self.bytes_read
+                     if self.limit is not None else self.limit)
             part = klass(self.fp, headers, ib, environ, keep_blank_values,
-                         strict_parsing,self.limit-self.bytes_read,
-                         self.encoding, self.errors)
+                         strict_parsing, limit, self.encoding, self.errors)
             self.bytes_read += part.bytes_read
             self.list.append(part)
             if part.done or self.bytes_read >= self.length > 0:
                 break
         self.skip_lines()
 
     def read_single(self):
         """Internal: read an atomic part."""
@@ -804,17 +805,17 @@ class FieldStorage:
         to bytes for comparisons.
         """
         next_boundary = b"--" + self.outerboundary
         last_boundary = next_boundary + b"--"
         delim = b""
         last_line_lfend = True
         _read = 0
         while 1:
-            if _read >= self.limit:
+            if self.limit is not None and _read >= self.limit:
                 break
             line = self.fp.readline(1<<16) # bytes
             self.bytes_read += len(line)
             _read += len(line)
             if not line:
                 self.done = -1
                 break
             if delim == b"\r":
diff --git a/Lib/test/test_cgi.py b/Lib/test/test_cgi.py
--- a/Lib/test/test_cgi.py
+++ b/Lib/test/test_cgi.py
@@ -243,16 +243,32 @@ class CgiTests(unittest.TestCase):
                   {'name':'title', 'filename':None, 'value':''},
                   {'name':'file', 'filename':'test.txt', 'value':b'Testing 123.\n'},
                   {'name':'submit', 'filename':None, 'value':' Add '}]
         for x in range(len(fs.list)):
             for k, exp in expect[x].items():
                 got = getattr(fs.list[x], k)
                 self.assertEqual(got, exp)
 
+    def test_fieldstorage_multipart_missing_content_length(self):
+        fp = BytesIO(b"""--MyBoundary
+Content-Disposition: form-data; name="my-arg"; filename="foo"
+
+Test
+
+--MyBoundary--
+""")
+        env = {
+            'REQUEST_METHOD': 'POST',
+            'CONTENT_TYPE': 'multipart/form-data; boundary=MyBoundary',
+            'wsgi.input': fp,
+        }
+        fs = cgi.FieldStorage(fp, environ=env)
+        self.assertEqual(len(fs['my-arg'].file.read()), 5)
+
     def test_fieldstorage_multipart_leading_whitespace(self):
         env = {
             'REQUEST_METHOD': 'POST',
             'CONTENT_TYPE': 'multipart/form-data; boundary={}'.format(BOUNDARY),
             'CONTENT_LENGTH': '560'}
         # Add some leading whitespace to our post data that will cause the
         # first line to not be the innerboundary.
         fp = BytesIO(b"\r\n" + POSTDATA.encode('latin-1'))