diff --git a/examples/har_dump.py b/examples/har_dump.py
new file mode 100644
index 000000000..0174da538
--- /dev/null
+++ b/examples/har_dump.py
@@ -0,0 +1,145 @@
+"""
+This inline script can be used to dump flows as HAR files.
+"""
+
+
+import pprint
+import json
+import sys
+
+from datetime import datetime
+import pytz
+
+import mitmproxy
+from mitmproxy import version
+
+HAR = {}
+
+
+def start():
+    """
+        Called once on script startup before any other events.
+    """
+    if len(sys.argv) != 2:
+        raise ValueError(
+            'Usage: -s "har_dump.py filename" '
+            '(- will output to stdout, filenames ending with .zhar '
+            'will result in compressed har)'
+        )
+
+    HAR.update({
+        "log": {
+            "version": "1.2",
+            "creator": {
+                "name": "mitmproxy har_dump",
+                "version": "0.1",
+                "comment": "mitmproxy version %s" % version.MITMPROXY
+            },
+            "pages": [],
+            "entries": []
+        }
+    })
+
+
+def response(flow):
+    """
+       Called when a server response has been received.
+    """
+    entries = HAR["log"]["entries"]
+
+    # TODO: SSL and Connect Timings
+
+    # Calculate raw timings from timestamps.
+
+    # DNS timings can not be calculated for lack of a way to measure it.
+    # The same goes for HAR blocked.
+
+    # mitmproxy will open a server connection as soon as it receives the host
+    # and port from the client connection. So, the time spent waiting is actually
+    # spent waiting between request.timestamp_end and response.timestamp_start thus it
+    # correlates to HAR wait instead.
+    timings_raw = {
+        'send': flow.request.timestamp_end - flow.request.timestamp_start,
+        'receive': flow.response.timestamp_end - flow.response.timestamp_start,
+        'wait': flow.response.timestamp_start - flow.request.timestamp_end,
+    }
+
+    # HAR timings are integers in ms, so we re-encode the raw timings to that format.
+    timings = dict([(k, int(1000 * v)) for k, v in timings_raw.items()])
+
+    # full_time is the sum of all timings.
+    # Timings set to -1 will be ignored as per spec.
+    full_time = sum(v for v in timings.values() if v > -1)
+
+    started_date_time = datetime.utcfromtimestamp(
+        flow.request.timestamp_start).replace(tzinfo=pytz.timezone("UTC")).isoformat()
+
+    # Size calculations
+    response_body_size = len(flow.response.content)
+    response_body_decoded_size = len(flow.response.content)
+    response_body_compression = response_body_decoded_size - response_body_size
+
+    entries.append({
+        "startedDateTime": started_date_time,
+        "time": full_time,
+        "request": {
+            "method": flow.request.method,
+            "url": flow.request.url,
+            "httpVersion": flow.request.http_version,
+            "cookies": name_value(flow.request.cookies),
+            "headers": name_value(flow.request.headers),
+            "queryString": name_value(flow.request.query or {}),
+            "headersSize": len(str(flow.request.headers)),
+            "bodySize": len(flow.request.content),
+        },
+        "response": {
+            "status": flow.response.status_code,
+            "statusText": flow.response.reason,
+            "httpVersion": flow.response.http_version,
+            "cookies": name_value(flow.response.cookies),
+            "headers": name_value(flow.response.headers),
+            "content": {
+                "size": response_body_size,
+                "compression": response_body_compression,
+                "mimeType": flow.response.headers.get('Content-Type', '')
+            },
+            "redirectURL": flow.response.headers.get('Location', ''),
+            "headersSize": len(str(flow.response.headers)),
+            "bodySize": response_body_size,
+        },
+        "cache": {},
+        "timings": timings,
+    })
+
+
+def done():
+    """
+        Called once on script shutdown, after any other events.
+    """
+    dump_file = sys.argv[1]
+
+    if dump_file == '-':
+        mitmproxy.ctx.log(pprint.pformat(HAR))
+    # TODO: .zhar compression
+    else:
+        with open(dump_file, "wb") as f:
+            f.write(json.dumps(HAR, indent=2))
+
+    # TODO: Log results via mitmproxy.ctx.log
+
+
+def name_value(obj):
+    """
+        Convert (key, value) pairs to HAR format.
+    """
+
+    items = []
+    if hasattr(obj, 'fields'):
+        items = obj.fields
+    elif hasattr(obj, 'items'):
+        items = obj.items()
+
+    if items:
+        return [{"name": k, "value": v} for k, v in items]
+    else:
+        return ""