diff --git a/ci/resources/supervisord/supervisord.yaml b/ci/resources/supervisord/supervisord.yaml
deleted file mode 100755
index 5bc09be5db..0000000000
--- a/ci/resources/supervisord/supervisord.yaml
+++ /dev/null
@@ -1,6 +0,0 @@
-init_config:
-
-instances:
-   - name: travis
-     socket: unix://VOLATILE_DIR//supervisor.sock
-     host: http://127.0.0.1
diff --git a/tests/checks/integration/test_supervisord.py b/tests/checks/integration/test_supervisord.py
index a6d09b9337..6a23f2866a 100644
--- a/tests/checks/integration/test_supervisord.py
+++ b/tests/checks/integration/test_supervisord.py
@@ -1,48 +1,87 @@
 # stdlib
-import os
 from time import sleep
-import unittest
 
 # 3p
 from nose.plugins.attrib import attr
 
 # project
-from tests.checks.common import get_check
+from checks import AgentCheck
+from tests.checks.common import AgentCheckTest
+
+PROCESSES = ["program_0", "program_1", "program_2"]
+STATUSES = ["down", "up", "unknown"]
 
 
 @attr(requires='supervisord')
-class TestSupervisordCheck(unittest.TestCase):
+class TestSupervisordCheck(AgentCheckTest):
+    CHECK_NAME = 'supervisord'
+
+    SUPERVISORD_CONFIG = [{
+        'name': "travis",
+        'socket': "unix://VOLATILE_DIR//supervisor.sock",
+    }]
 
-    def test_travis_supervisord(self):
-        """Integration test for supervisord check. Using a supervisord on Travis."""
+    BAD_SUPERVISORD_CONFIG = [{
+        'name': "travis",
+        'socket': "unix:///wrong/path/supervisor.sock",
+    }]
 
-        # Load yaml config
-        config_str = open(os.environ['VOLATILE_DIR'] + '/supervisor/supervisord.yaml', 'r').read()
-        self.assertTrue(config_str is not None and len(config_str) > 0, msg=config_str)
+    # Supervisord should run 3 programs for 10, 20 and 30 seconds
+    # respectively.
+    # The following dictionnary shows the processes by state for each iteration.
+    PROCESSES_BY_STATE_BY_ITERATION = map(
+        lambda x: dict(up=PROCESSES[x:], down=PROCESSES[:x], unknown=[]),
+        range(4)
+    )
 
-        # init the check and get the instances
-        check, instances = get_check('supervisord', config_str)
-        self.assertTrue(check is not None, msg=check)
-        self.assertEquals(len(instances), 1)
+    def test_check(self):
+        """
+        Run Supervisord check and assess coverage
+        """
+        config = {'instances': self.SUPERVISORD_CONFIG}
+        instance_tags = ["supervisord_server:travis"]
 
-        # Supervisord should run 3 programs for 30, 60 and 90 seconds
-        # respectively. The tests below will ensure that the process count
-        # metric is reported correctly after (roughly) 10, 40, 70 and 100 seconds
         for i in range(4):
-            try:
-                # Run the check
-                check.check(instances[0])
-            except Exception, e:
-                # Make sure that it ran successfully
-                self.assertTrue(False, msg=str(e))
-            else:
-                up, down = 0, 0
-                for name, timestamp, value, meta in check.get_metrics():
-                    if name == 'supervisord.process.count':
-                        if 'status:up' in meta['tags']:
-                            up = value
-                        elif 'status:down' in meta['tags']:
-                            down = value
-                self.assertEquals(up, 3 - i)
-                self.assertEquals(down, i)
-                sleep(10)
+            # Run the check
+            self.run_check(config)
+
+            # Check metrics and service checks scoped by process
+            for proc in PROCESSES:
+                process_tags = instance_tags + ["supervisord_process:{0}".format(proc)]
+                process_status = AgentCheck.OK if proc in \
+                    self.PROCESSES_BY_STATE_BY_ITERATION[i]['up'] else AgentCheck.CRITICAL
+
+                self.assertMetric("supervisord.process.uptime", tags=process_tags, count=1)
+                self.assertServiceCheck("supervisord.process.status", status=process_status,
+                                        tags=process_tags, count=1)
+            # Check instance metrics
+            for status in STATUSES:
+                status_tags = instance_tags + ["status:{0}".format(status)]
+                count_processes = len(self.PROCESSES_BY_STATE_BY_ITERATION[i][status])
+                self.assertMetric("supervisord.process.count", value=count_processes,
+                                  tags=status_tags, count=1)
+
+            # Check service checks
+            self.assertServiceCheck("supervisord.can_connect", status=AgentCheck.OK,
+                                    tags=instance_tags, count=1)
+
+            # Raises when COVERAGE=true and coverage < 100%
+            self.coverage_report()
+
+            # Sleep 10s to give enough time to processes to terminate
+            sleep(10)
+
+    def test_connection_falure(self):
+        """
+        Service check reports connection failure
+        """
+        config = {'instances': self.BAD_SUPERVISORD_CONFIG}
+        instance_tags = ["supervisord_server:travis"]
+
+        self.assertRaises(
+            Exception,
+            lambda: self.run_check(config)
+        )
+        self.assertServiceCheck("supervisord.can_connect", status=AgentCheck.CRITICAL,
+                                tags=instance_tags, count=1)
+        self.coverage_report()