diff --git a/CHANGES.rst b/CHANGES.rst
index 9877c43..5a203ec 100644
--- a/CHANGES.rst
+++ b/CHANGES.rst
@@ -1,11 +1,12 @@
 Version History
 ===============================================================================
 
-Version: 4.0.3
+Version: 4.0.4
 -------------------------------------------------------------------------------
 ENHANCEMENTS:
 
 * Added the 'worker_id' context.config.userdata parameter to allow users to identify which worker is executing every feature or scenario when running tests in parallel. `PR #121 <https://github.com/hrcorval/behavex/pull/121>`_
+* Adding the --parallel-delay argument, to enable setting a staggered execution when running tests in parallel. `Issue #142 <https://github.com/hrcorval/behavex/issues/142>`_
 
 CONTRIBUTIONS:
 
diff --git a/behavex/arguments.py b/behavex/arguments.py
index 333dfd7..5d4feef 100644
--- a/behavex/arguments.py
+++ b/behavex/arguments.py
@@ -285,6 +285,11 @@ def parse_arguments(args):
         help="Specifies whether parallel execution should be performed at the scenario or feature level.",
         required=False,
     )
+    parser.add_argument(
+        '--parallel-delay',
+        type=int,
+        default=0,
+        help='Delay in milliseconds before starting each parallel process')
     parser.add_argument(
         '-ip',
         '--include-paths',
diff --git a/behavex/runner.py b/behavex/runner.py
index 631f0c9..50584e2 100644
--- a/behavex/runner.py
+++ b/behavex/runner.py
@@ -167,13 +167,16 @@ def setup_running_failures(args_parsed):
         return EXIT_OK, None
 
 
-def init_multiprocessing(idQueue):
+def init_multiprocessing(idQueue, parallel_delay):
     """Initialize multiprocessing by ignoring SIGINT signals."""
     signal.signal(signal.SIGINT, signal.SIG_IGN)
     # Retrieve one of the unique IDs
     worker_id = idQueue.get()
     # Use the unique ID to name the process
     multiprocessing.current_process().name = f'behave_worker-{worker_id}'
+    # Add delay
+    if parallel_delay > 0:
+        time.sleep(parallel_delay / 1000.0)
 
 
 def launch_behavex():
@@ -217,9 +220,10 @@ def launch_behavex():
     idQueue = manager.Queue()
     for i in range(parallel_processes):
         idQueue.put(i)
+    parallel_delay = get_param('parallel_delay')
     process_pool = ProcessPoolExecutor(max_workers=parallel_processes,
                                        initializer=init_multiprocessing,
-                                       initargs=(idQueue,))
+                                       initargs=(idQueue, parallel_delay))
     global_vars.execution_start_time = time.time()
     try:
         config = ConfigRun()
diff --git a/setup.py b/setup.py
index 249bc91..82ff1c1 100644
--- a/setup.py
+++ b/setup.py
@@ -6,7 +6,7 @@
 
 setup(
     name='behavex',
-    version='4.0.2',
+    version='4.0.4rc1',
     license="MIT",
     platforms=['any'],
     python_requires='>=3.5',