diff --git a/src/testdir/test_plus_arg_edit.vim b/src/testdir/test_plus_arg_edit.vim
index 71dbea1..e91a6e4 100644
--- a/src/testdir/test_plus_arg_edit.vim
+++ b/src/testdir/test_plus_arg_edit.vim
@@ -1,10 +1,38 @@
 " Tests for complicated + argument to :edit command
 function Test_edit()
-  call writefile(["foo|bar"], "Xfile1") 
-  call writefile(["foo/bar"], "Xfile2") 
+  call writefile(["foo|bar"], "Xfile1")
+  call writefile(["foo/bar"], "Xfile2")
   edit +1|s/|/PIPE/|w Xfile1| e Xfile2|1 | s/\//SLASH/|w
   call assert_equal(["fooPIPEbar"], readfile("Xfile1"))
   call assert_equal(["fooSLASHbar"], readfile("Xfile2"))
   call delete('Xfile1')
   call delete('Xfile2')
 endfunction
+
+func Test_edit_bad()
+  if !has('multi_byte')
+    finish
+  endif
+
+  " Test loading a utf8 file with bad utf8 sequences.
+  call writefile(["[\xff][\xc0][\xe2\x89\xf0][\xc2\xc2]"], "Xfile")
+  new
+
+  " Without ++bad=..., the default behavior is like ++bad=?
+  e! ++enc=utf8 Xfile
+  call assert_equal('[?][?][???][??]', getline(1))
+
+  e! ++enc=utf8 ++bad=_ Xfile
+  call assert_equal('[_][_][___][__]', getline(1))
+
+  e! ++enc=utf8 ++bad=drop Xfile
+  call assert_equal('[][][][]', getline(1))
+
+  e! ++enc=utf8 ++bad=keep Xfile
+  call assert_equal("[\xff][\xc0][\xe2\x89\xf0][\xc2\xc2]", getline(1))
+
+  call assert_fails('e! ++enc=utf8 ++bad=foo Xfile', 'E474:')
+
+  bw!
+  call delete('Xfile')
+endfunc