diff --git a/src/calibre/ebooks/conversion/plugins/txt_input.py b/src/calibre/ebooks/conversion/plugins/txt_input.py
index 4f73a067a6..5cab2a1a96 100644
--- a/src/calibre/ebooks/conversion/plugins/txt_input.py
+++ b/src/calibre/ebooks/conversion/plugins/txt_input.py
@@ -236,5 +236,14 @@ class TXTInput(InputFormatPlugin):
         from calibre.ebooks.oeb.transforms.metadata import meta_info_to_oeb_metadata
         mi = get_file_type_metadata(stream, file_ext)
         meta_info_to_oeb_metadata(mi, oeb.metadata, log)
+        self.html_postprocess_title = mi.title
 
         return oeb
+
+    def postprocess_book(self, oeb, opts, log):
+        for item in oeb.spine:
+            if hasattr(item.data, 'xpath'):
+                for title in item.data.xpath('//*[local-name()="title"]'):
+                    if title.text == _('Unknown'):
+                        title.text = self.html_postprocess_title
+
diff --git a/src/calibre/ebooks/metadata/txt.py b/src/calibre/ebooks/metadata/txt.py
index 70d3c72ae0..9140fb6f61 100644
--- a/src/calibre/ebooks/metadata/txt.py
+++ b/src/calibre/ebooks/metadata/txt.py
@@ -7,7 +7,7 @@ __copyright__ = '2009, John Schember <john@nachtimwald.com>'
 Read meta information from TXT files
 '''
 
-import re
+import re, os
 
 from calibre.ebooks.metadata import MetaInformation
 
@@ -15,7 +15,10 @@ def get_metadata(stream, extract_cover=True):
     '''
     Return metadata as a L{MetaInfo} object
     '''
-    mi = MetaInformation(_('Unknown'), [_('Unknown')])
+    name = getattr(stream, 'name', '').rpartition('.')[0]
+    if name:
+        name = os.path.basename(name)
+    mi = MetaInformation(name or _('Unknown'), [_('Unknown')])
     stream.seek(0)
 
     mdata = u''
@@ -29,7 +32,7 @@ def get_metadata(stream, extract_cover=True):
     mdata = mdata[:100]
 
     mo = re.search('(?u)^[ ]*(?P<title>.+)[ ]*(\n{3}|(\r\n){3}|\r{3})[ ]*(?P<author>.+)[ ]*(\n|\r\n|\r)$', mdata)
-    if mo != None:
+    if mo is not None:
         mi.title = mo.group('title')
         mi.authors = mo.group('author').split(',')