diff --git a/swh/scheduler/cli/simulator.py b/swh/scheduler/cli/simulator.py
--- a/swh/scheduler/cli/simulator.py
+++ b/swh/scheduler/cli/simulator.py
@@ -57,8 +57,11 @@
     "showplots",
     help="Show results as plots (with plotille)",
 )
+@click.option(
+    "--csv", "-o", "csvfile", type=click.File("w"), help="Export results in a CSV file"
+)
 @click.pass_context
-def run_command(ctx, scheduler, policy, runtime, showplots):
+def run_command(ctx, scheduler, policy, runtime, showplots, csvfile):
     """Run the scheduler simulator.
 
     By default, the simulation runs forever. You can cap the simulated runtime
@@ -80,3 +83,5 @@
     )
 
     print(report.format(with_plots=showplots))
+    if csvfile is not None:
+        report.metrics_csv(csvfile)
diff --git a/swh/scheduler/simulator/common.py b/swh/scheduler/simulator/common.py
--- a/swh/scheduler/simulator/common.py
+++ b/swh/scheduler/simulator/common.py
@@ -3,10 +3,11 @@
 # License: GNU General Public License version 3, or any later version
 # See top-level LICENSE file for more information
 
+import csv
 from dataclasses import dataclass, field
 from datetime import datetime, timedelta
 import textwrap
-from typing import Dict, List, Optional, Tuple
+from typing import Dict, List, Optional, TextIO, Tuple
 import uuid
 
 import plotille
@@ -32,7 +33,11 @@
     scheduler_metrics: List[Tuple[datetime, List[SchedulerMetrics]]] = field(
         default_factory=list
     )
-    """Collected scheduler metrics for every timestamp"""
+    """Collected scheduler metrics
+
+    This is a list of couples (timestamp, [SchedulerMetrics,]): the list of
+    scheduler metrics collected at given timestamp.
+    """
 
     visit_metrics: List[Tuple[datetime, int]] = field(default_factory=list)
     """Collected visit metrics over time"""
@@ -90,6 +95,33 @@
 
         return figure.show(legend=True)
 
+    def metrics_csv(self, fobj: TextIO) -> None:
+        """Export scheduling metrics in a csv file"""
+        csv_writer = csv.writer(fobj)
+        csv_writer.writerow(
+            [
+                "timestamp",
+                "known_origins",
+                "enabled_origins",
+                "never_visited_origins",
+                "origins_with_pending_changes",
+            ]
+        )
+
+        timestamps, metric_lists = zip(*self.scheduler_metrics)
+        known = (sum(m.origins_known for m in metrics) for metrics in metric_lists)
+        enabled = (sum(m.origins_enabled for m in metrics) for metrics in metric_lists)
+        never_visited = (
+            sum(m.origins_never_visited for m in metrics) for metrics in metric_lists
+        )
+        pending_changes = (
+            sum(m.origins_with_pending_changes for m in metrics)
+            for metrics in metric_lists
+        )
+        csv_writer.writerows(
+            zip(timestamps, known, enabled, never_visited, pending_changes)
+        )
+
     def format(self, with_plots=True):
         full_visits = self.visit_runtimes.get(("full", True), [])
         long_tasks = sum(runtime > self.DURATION_THRESHOLD for runtime in full_visits)
diff --git a/swh/scheduler/tests/test_simulator.py b/swh/scheduler/tests/test_simulator.py
--- a/swh/scheduler/tests/test_simulator.py
+++ b/swh/scheduler/tests/test_simulator.py
@@ -3,6 +3,8 @@
 # License: GNU General Public License version 3, or any later version
 # See top-level LICENSE file for more information
 
+import io
+
 import pytest
 
 from swh.core.api.classes import stream_results
@@ -62,3 +64,6 @@
     # just check these SimulationReport methods do not crash
     assert report.format(with_plots=True)
     assert report.format(with_plots=False)
+    fobj = io.StringIO()
+    report.metrics_csv(fobj=fobj)
+    assert fobj.getvalue()