diff --git a/dlibrary/dlibrary.py b/dlibrary/dlibrary.py
index 1a7824d..4c9b210 100755
--- a/dlibrary/dlibrary.py
+++ b/dlibrary/dlibrary.py
@@ -29,7 +29,8 @@ FAKKU_ID_REGEX = re.compile('.*_FAKKU$')
 TEXTLESS_REGEX = re.compile('(台詞|セリフ|せりふ|テキスト|文字)((な|無)し|抜き)|notext|textless', re.IGNORECASE)
 EPILOGUE_REGEX = re.compile('after|後日談', re.IGNORECASE)
 HI_RES_REGEX = re.compile('高解像度', re.IGNORECASE)
-COVER_REGEX = re.compile('表紙|cover|hyoushi', re.IGNORECASE)
+FRONT_COVER_REGEX = re.compile('(^|[^裏])表紙|cover|hyoushi', re.IGNORECASE)
+BACK_COVER_REGEX = re.compile('裏表紙', re.IGNORECASE)
 ALT_VERSIONS = [
     '褐色',
     '日焼け',
@@ -525,14 +526,14 @@ def collate_from_paths(srcs, dest, start_index, exclude):
     if textless_split != False:
         return textless_split
 
+    cover_split = try_collate_split_regex(srcs, dest, start_index, exclude, earlier=FRONT_COVER_REGEX, later=BACK_COVER_REGEX)
+    if cover_split != False:
+        return cover_split
+
     epilogue_split = try_collate_split_regex(srcs, dest, start_index, exclude, later=EPILOGUE_REGEX)
     if epilogue_split != False:
         return epilogue_split
 
-    cover_split = try_collate_split_regex(srcs, dest, start_index, exclude, earlier=COVER_REGEX)
-    if cover_split != False:
-        return cover_split
-
     if all(src.is_file() and is_image(src) for src in srcs):
         ordering = complete_prefix_number_ordering(srcs)
         if ordering: