diff --git a/rally/benchmark/sla/base.py b/rally/benchmark/sla/base.py
index 98e8df762f..a46ae30aa0 100644
--- a/rally/benchmark/sla/base.py
+++ b/rally/benchmark/sla/base.py
@@ -24,9 +24,17 @@ import abc
 import jsonschema
 import six
 
+from rally.openstack.common.gettextutils import _
 from rally import utils
 
 
+class SLAResult(object):
+
+    def __init__(self, success=True, msg=None):
+        self.success = success
+        self.msg = msg
+
+
 @six.add_metaclass(abc.ABCMeta)
 class SLA(object):
     """Factory for criteria classes."""
@@ -67,11 +75,12 @@ class SLA(object):
             config = result['key']['kw'].get('sla', None)
             if config:
                 for name, criterion in config.iteritems():
-                    success = opt_name_map[name].check(criterion, result)
+                    check_result = opt_name_map[name].check(criterion, result)
                     yield {'benchmark': result['key']['name'],
                            'pos': result['key']['pos'],
                            'criterion': name,
-                           'success': success}
+                           'success': check_result.success,
+                           'detail': check_result.msg}
 
 
 class FailureRate(SLA):
@@ -84,8 +93,12 @@ class FailureRate(SLA):
         raw = result['data']['raw']
         errors = len(filter(lambda x: x['error'], raw))
         if criterion_value < errors * 100.0 / len(raw):
-            return False
-        return True
+            success = False
+        else:
+            success = True
+        msg = (_("Maximum failure percent %s%% failures, actually %s%%") %
+                (criterion_value, errors * 100.0 / len(raw)))
+        return SLAResult(success, msg)
 
 
 class IterationTime(SLA):
@@ -96,7 +109,13 @@ class IterationTime(SLA):
 
     @staticmethod
     def check(criterion_value, result):
+        duration = 0
+        success = True
         for i in result['data']['raw']:
+            duration = i['duration']
             if i['duration'] > criterion_value:
-                return False
-        return True
+                success = False
+                break
+        msg = (_("Maximum seconds per iteration %is, actually %is") %
+                (criterion_value, duration))
+        return SLAResult(success, msg)
diff --git a/rally/cmd/commands/task.py b/rally/cmd/commands/task.py
index b4d7965b88..13692900d3 100644
--- a/rally/cmd/commands/task.py
+++ b/rally/cmd/commands/task.py
@@ -388,5 +388,6 @@ class TaskCommands(object):
             print(json.dumps(rows))
         else:
             common_cliutils.print_list(rows, ('benchmark', 'pos',
-                                              'criterion', 'success'))
+                                              'criterion', 'success',
+                                              'detail'))
         return failed_criteria
diff --git a/tests/benchmark/sla/test_base.py b/tests/benchmark/sla/test_base.py
index c3317d22bc..634bc57d30 100644
--- a/tests/benchmark/sla/test_base.py
+++ b/tests/benchmark/sla/test_base.py
@@ -27,7 +27,8 @@ class TestCriterion(base.SLA):
 
     @staticmethod
     def check(criterion_value, result):
-        return criterion_value == result["data"]
+        return base.SLAResult(criterion_value == result["data"],
+                              msg='detail')
 
 
 class BaseSLATestCase(test.TestCase):
@@ -54,6 +55,7 @@ class BaseSLATestCase(test.TestCase):
         results = list(base.SLA.check_all(task))
         expected = [{'benchmark': 'fake',
                      'criterion': 'test_criterion',
+                     'detail': 'detail',
                      'pos': 0,
                      'success': True}]
         self.assertEqual(expected, results)
@@ -61,6 +63,7 @@ class BaseSLATestCase(test.TestCase):
         results = list(base.SLA.check_all(task))
         expected = [{'benchmark': 'fake',
                      'criterion': 'test_criterion',
+                     'detail': 'detail',
                      'pos': 0,
                      'success': False}]
         self.assertEqual(expected, results)
@@ -73,8 +76,10 @@ class FailureRateTestCase(test.TestCase):
                 {"error": []},
         ]  # one error and one success. 50% success rate
         result = {"data": {"raw": raw}}
-        self.assertTrue(base.FailureRate.check(75.0, result))  # 50% < 75.0%
-        self.assertFalse(base.FailureRate.check(25, result))  # 50% > 25%
+        # 50% < 75.0%
+        self.assertTrue(base.FailureRate.check(75.0, result).success)
+        # 50% > 25%
+        self.assertFalse(base.FailureRate.check(25, result).success)
 
 
 class IterationTimeTestCase(test.TestCase):
@@ -84,5 +89,5 @@ class IterationTimeTestCase(test.TestCase):
                 {"duration": 6.28},
         ]
         result = {"data": {"raw": raw}}
-        self.assertTrue(base.IterationTime.check(42, result))
-        self.assertFalse(base.IterationTime.check(3.62, result))
+        self.assertTrue(base.IterationTime.check(42, result).success)
+        self.assertFalse(base.IterationTime.check(3.62, result).success)