diff --git a/modules/okc/backend.py b/modules/okc/backend.py
index a1d94693..03a38694 100644
--- a/modules/okc/backend.py
+++ b/modules/okc/backend.py
@@ -218,8 +218,16 @@ class OkCBackend(BaseBackend, ICapMessages, ICapContact, ICapMessagesPost):
     # ---- ICapMessagesPost methods ---------------------
 
     def post_message(self, message):
+        content = message.content.replace('\n', '\r\n').encode('utf-8', 'replace')
         with self.browser:
-            self.browser.post_mail(message.thread.id, message.content)
+            # Check wether we already have a thread with this user
+            threads = self.browser.get_threads_list()
+            for thread in threads:
+                if thread['username'] == message.thread.id:
+                    self.browser.post_reply(thread['id'], content)
+                    break
+            else:
+                self.browser.post_mail(message.thread.id, content)
 
     # ---- ICapContact methods ---------------------
 
diff --git a/modules/okc/browser.py b/modules/okc/browser.py
index cec6b72f..061d9dc8 100644
--- a/modules/okc/browser.py
+++ b/modules/okc/browser.py
@@ -17,6 +17,7 @@
 # You should have received a copy of the GNU Affero General Public License
 # along with weboob. If not, see <http://www.gnu.org/licenses/>.
 
+import urllib
 
 from weboob.tools.browser import BaseBrowser
 from weboob.tools.ordereddict import OrderedDict
@@ -144,9 +145,25 @@ class OkCBrowser(BaseBrowser):
     @check_login
     def post_mail(self, id, content):
         self.location(self.absurl('/messages?compose=1'))
-        content = content.replace('\n', '\r\n').encode('Windows-1252', 'replace')
         self.page.post_mail(id, content)
 
+    @check_login
+    def post_reply(self, thread_id, content):
+        self.location(self.absurl('/messages?readmsg=true&threadid=%s&folder=1' % thread_id))
+        username, key = self.page.get_post_params()
+        data = urllib.urlencode({
+            'ajax' : 1,
+            'sendmsg' : 1,
+            'r1' : username,
+            'subject' : '',
+            'body' : content,
+            'threadid' : thread_id,
+            'authcode' : key,
+            'reply' : 1,
+            })
+        self.addheaders = [('Referer', self.page.url), ('Content-Type', 'application/x-www-form-urlencoded')]
+        self.open('http://m.okcupid.com/mailbox', data=data)
+
     #@check_login
     #@url2id
     #def delete_thread(self, id):
diff --git a/modules/okc/pages.py b/modules/okc/pages.py
index e3a9500f..23f8f37c 100644
--- a/modules/okc/pages.py
+++ b/modules/okc/pages.py
@@ -17,6 +17,7 @@
 # You should have received a copy of the GNU Affero General Public License
 # along with weboob. If not, see <http://www.gnu.org/licenses/>.
 
+import re
 
 from weboob.tools.browser import BasePage
 from weboob.tools.ordereddict import OrderedDict
@@ -67,6 +68,31 @@ class MessagesPage(BasePage):
 
         return mails
 
+    def get_post_params(self):
+        # http://m.okcupid.com/mailbox
+        # Paramètresapplication/x-www-form-urlencoded
+        # ajax    1
+        # authcode    1,0,1332766806,0x154df106d5af5993;51e1fa019f3423831e5b95c9c91346e5138f99cf
+        # body    Ah bah ça y'est, ça marche ?!
+        # r1  jeanneplop
+        # reply   1
+        # sendmsg 1
+        # subject
+        # threadid    154028265985080112
+        # ajax=1&sendmsg=1&r1=jeanneplop&subject=&body=Ah%20bah%20%C3%A7a%20y'est%2C%20%C3%A7a%20marche%20%3F!&threadid=154028265985080112&authcode=1%2C0%2C1332766806%2C0x154df106d5af5993%3B51e1fa019f3423831e5b95c9c91346e5138f99cf&reply=1
+        js = self.parser.select(self.document.getroot(), "//script", method='xpath')
+        for script in js:
+            script = script.text
+
+            if script is None:
+                continue
+            for line in script.splitlines():
+                match = re.match(".*Message\.initialize\([^,]*, '([^']*)', \"\", '([^']*)',.*", line)
+                if match is not None:
+                    return match.groups()
+        raise Exception('Unexpected reply page')
+
+
 class ProfilePage(BasePage):
     def get_profile(self):
         title = self.parser.select(self.document.getroot(), 'title', 1)