diff --git a/provingground/cros_perf/perf.json b/provingground/cros_perf/perf.json
new file mode 100644
index 0000000..4cc585c
--- /dev/null
+++ b/provingground/cros_perf/perf.json
@@ -0,0 +1,62 @@
+{
+    "bots": [
+        {"device_name": "cros-alex"},
+        {"device_name": "cros-snow"},
+        {"device_name": "cros-parrot"},
+        {"device_name": "cros-peppy"},
+        {"device_name": "cros-link"},
+        {"device_name": "cros-peach_pit"},
+        {"device_name": "cros-squawks"},
+        {"device_name": "cros-nyan_big"}
+    ],
+    "tests": [
+        {
+            "test_name": "platform_BootPerfServer/seconds_power_on_to_login",
+            "masters": "ChromeOSPerf"
+        },
+        {
+            "test_name": "platform_BootPerfServer/seconds_shutdown_time",
+            "masters": "ChromeOSPerf"
+        },
+        {
+            "test_name": "graphics_WebGLManyPlanetsDeep/average_fps",
+            "masters": "ChromeOS_Graphics"
+        },
+        {
+            "test_name": "graphics_WebGLAquarium/avg_fps_1000_fishes",
+            "masters": "ChromeOS_Graphics"
+        },
+        {
+            "test_name": "graphics_WebGLPerformance/time_geom_mean/time_geom_mean",
+            "masters": "ChromeOS_Graphics"
+        },
+        {
+            "test_name": "video_VDAPerf/crowd1080_h264.delivery_time",
+            "masters": "ChromeOSVideo"
+        },
+        {
+            "test_name": "video_VDAPerf/crowd1080_h264.decode_time.percentile_0.50",
+            "masters": "ChromeOSVideo"
+        },
+        {
+            "test_name": "video_WebRtcPerf/hw_decode_time.percentile_0.50",
+            "masters": "ChromeOSVideo"
+        },
+        {
+            "test_name": "video_WebRtcPerf/hw_decode_time.max",
+            "masters": "ChromeOSVideo"
+        },
+        {
+            "test_name": "power_Resume/system_suspend",
+            "masters": "ChromeOSPower"
+        },
+        {
+            "test_name": "power_Resume/system_resume",
+            "masters": "ChromeOSPower"
+        },
+        {
+            "test_name": "platform_Pkcs11LoadPerf/Key_Ready",
+            "masters": "ChromeOSPerf"
+        }
+    ]
+}
diff --git a/provingground/cros_perf/perf_results.py b/provingground/cros_perf/perf_results.py
new file mode 100644
index 0000000..b536bf8
--- /dev/null
+++ b/provingground/cros_perf/perf_results.py
@@ -0,0 +1,132 @@
+#!/usr/bin/python
+# Copyright 2014 The Chromium OS Authors. All rights reserved.
+# Use of this source code is governed by a BSD-style license that can be
+# found in the LICENSE file.
+
+import argparse
+import csv
+from io import BytesIO
+import json
+import pycurl
+from urllib import urlencode
+
+
+_PERF_DASHBOARD_URL = 'https://chromeperf.appspot.com/graph_json'
+_PERF_RESULT_FILE = 'mstone.csv'
+_VALUE_ERROR = -1
+
+
+def main():
+    """Retrieves perf data from Chrome perf dashboard.
+
+    Script to retrieve performance data from the Chrome perf dashboard. Test
+    and bots configuration needs to be stored in a json file with expected
+    format. Check perf.json for a sample config file. On a successful data
+    retrieval, the results will be stored in mston.csv file which can be
+    imported in Google Sheet.
+    """
+    usage_desc = ("""
+         This script retrieves ChromeOS performance data from Chrome perf
+         dashboard. Device and test configuration should be stored into a json
+         file in the expected format.
+         e.g.
+         python perf_results.py -c perf.json -s 2171026063 -e 10360631002200
+         Revision numbers can be captured from the perf dahsboard by hovering
+         on perf data points.""")
+    parser = argparse.ArgumentParser(description=usage_desc)
+    parser.add_argument('-c', '--config', dest='perf_config_file',
+                        help='Json file with device and test config.')
+    parser.add_argument('-s', '--start_rev', dest='start_rev',
+                        help='Start revision point from the dashboard.')
+    parser.add_argument('-e', '--end_rev', dest='end_rev',
+                        help='End revision point from the dashboard.')
+    arguments = parser.parse_args()
+
+    if not arguments.perf_config_file:
+        parser.error('Config file is missing.')
+    if not arguments.start_rev or not arguments.start_rev.isdigit():
+        parser.error('Start revision is missing or not valid.')
+    if not arguments.end_rev or not arguments.end_rev.isdigit():
+        parser.error('End revision is missing or not valid.')
+
+    try:
+        with open(arguments.perf_config_file) as f:
+            conf = json.loads(f.read())
+    except IOError:
+        print 'Config file is not found.'
+        return
+
+    start_rev = arguments.start_rev
+    end_rev = arguments.end_rev
+
+    bots = [conf['bots'][i]['device_name'] for i in range(len(conf['bots']))]
+    tests = conf['tests']
+    curl_request = init_curl()
+    curl_buffer = BytesIO()
+
+    result_file = open(_PERF_RESULT_FILE, 'w')
+    try:
+        writer = csv.writer(result_file)
+        writer.writerow([' '] + bots)
+        for test in tests:
+            test_results = []
+            test_results.append(test['test_name'])
+            for bot in bots:
+                postfields = construct_post_data(bot, test, start_rev, end_rev)
+                mean = process_curl_query(curl_buffer, curl_request, postfields)
+                if mean == _VALUE_ERROR:
+                    test_results.append('ERR')
+                else:
+                    test_results.append(round(mean, 2))
+            writer.writerow(test_results)
+    finally:
+        curl_request.close()
+        result_file.close()
+
+
+def init_curl():
+    c = pycurl.Curl()
+    c.setopt(pycurl.URL, _PERF_DASHBOARD_URL)
+    return c
+
+
+def process_curl_query(curl_buffer, curl_request, postfields):
+    """Executes the curl request and generates mean for the test/bot."""
+    curl_request.setopt(curl_request.POSTFIELDS, postfields)
+    curl_request.setopt(curl_request.WRITEFUNCTION, curl_buffer.write)
+    curl_request.perform()
+    packet = curl_buffer.getvalue()
+    curl_buffer.truncate(0)
+    curl_buffer.seek(0)
+    return packet_to_mean(packet)
+
+
+def packet_to_mean(packet):
+    """Extracts perf values and returns the mean."""
+    try:
+        values_set = json.loads(packet)['data']
+    except ValueError:
+        return _VALUE_ERROR
+
+    if (len(values_set) == 0) or \
+            (len(values_set[0]) == 0) or \
+            (len(values_set[0]['data']) == 0):
+        return _VALUE_ERROR
+    values = values_set[0]['data']
+    return sum([values[i][1] for i in range(len(values))]) / len(values)
+
+
+def construct_post_data(bot, test, start_rev, end_rev):
+    graphs = {
+        "masters": [test['masters']],
+        "bots": [bot],
+        "tests": [test['test_name']],
+        "start_rev": start_rev,
+        "end_rev": end_rev,
+    }
+    post_data = {"graphs": json.dumps(graphs)}
+    return urlencode(post_data)
+
+
+if __name__ == "__main__":
+    main()