diff --git a/jobs/director/spec b/jobs/director/spec
index a7c88883521..038cb9d1fcc 100644
--- a/jobs/director/spec
+++ b/jobs/director/spec
@@ -282,6 +282,12 @@ properties:
     description: Client certificate for mutual TLS connections to an external metrics server
   director.metrics_server.tls.private_key:
     description: Client private key for mutual TLS connections to an external metrics server
+  director.metrics_server.file_retention_days:
+    description: 'Number of days to retain metric binary files in /var/vcap/store/director/metrics. Older files are automatically cleaned up by a scheduled job. Set to 0 to disable cleanup.'
+    default: 7
+  director.metrics_server.cleanup_schedule:
+    description: 'RufusScheduler cron formatted schedule for cleanup of stale metrics files'
+    default: '0 0 0 * * * UTC'  # once every day at midnight UTC
 
   # NATs
   nats.address:
diff --git a/jobs/director/templates/director.yml.erb b/jobs/director/templates/director.yml.erb
index 42aa781702a..9a8aa51f74f 100644
--- a/jobs/director/templates/director.yml.erb
+++ b/jobs/director/templates/director.yml.erb
@@ -162,6 +162,16 @@ params['scheduled_jobs'] << {
   'schedule' => p('director.tasks_cleanup_schedule')
 }
 
+if p('director.metrics_server.enabled')
+  params['scheduled_jobs'] << {
+    'command' => 'ScheduledMetricsCleanup',
+    'schedule' => p('director.metrics_server.cleanup_schedule'),
+    'params' => [{
+      'retention_days' => p('director.metrics_server.file_retention_days')
+    }]
+  }
+end
+
 params['record_events'] = p('director.events.record_events')
 if params['record_events']
   params['scheduled_jobs'] << {
diff --git a/src/bosh-director/bin/bosh-director b/src/bosh-director/bin/bosh-director
index 536a1bc16e4..c5e06970894 100755
--- a/src/bosh-director/bin/bosh-director
+++ b/src/bosh-director/bin/bosh-director
@@ -29,7 +29,7 @@ route_configuration = Bosh::Director::Api::RouteConfiguration.new(config)
 rack_app = Puma::Rack::Builder.app do
   use Rack::CommonLogger
   if config.metrics_server_enabled
-    Prometheus::Client.config.data_store = Prometheus::Client::DataStores::DirectFileStore.new(dir: '/var/vcap/store/director/metrics')
+    Prometheus::Client.config.data_store = Prometheus::Client::DataStores::DirectFileStore.new(dir: config.metrics_dir)
     use Bosh::Director::StripDeploymentsMiddlewareCollector
     use Prometheus::Middleware::Exporter
   end
diff --git a/src/bosh-director/lib/bosh/director.rb b/src/bosh-director/lib/bosh/director.rb
index 8696bb2d36a..7b6f150fefd 100644
--- a/src/bosh-director/lib/bosh/director.rb
+++ b/src/bosh-director/lib/bosh/director.rb
@@ -197,6 +197,7 @@ module Director
 require 'bosh/director/jobs/scheduled_dns_blobs_cleanup'
 require 'bosh/director/jobs/scheduled_dns_tombstone_cleanup'
 require 'bosh/director/jobs/scheduled_tasks_cleanup'
+require 'bosh/director/jobs/scheduled_metrics_cleanup'
 require 'bosh/director/jobs/create_snapshot'
 require 'bosh/director/jobs/snapshot_deployment'
 require 'bosh/director/jobs/snapshot_deployments'
diff --git a/src/bosh-director/lib/bosh/director/config.rb b/src/bosh-director/lib/bosh/director/config.rb
index 25a99f281b3..6fd9f9e0b0e 100644
--- a/src/bosh-director/lib/bosh/director/config.rb
+++ b/src/bosh-director/lib/bosh/director/config.rb
@@ -265,6 +265,10 @@ def log_dir
         File.dirname(@log_file_path) if @log_file_path
       end
 
+      def metrics_dir
+        File.join(@base_dir, 'metrics')
+      end
+
       def local_dns_enabled?
         !!@local_dns_enabled
       end
diff --git a/src/bosh-director/lib/bosh/director/jobs/scheduled_metrics_cleanup.rb b/src/bosh-director/lib/bosh/director/jobs/scheduled_metrics_cleanup.rb
new file mode 100644
index 00000000000..447bfdc4859
--- /dev/null
+++ b/src/bosh-director/lib/bosh/director/jobs/scheduled_metrics_cleanup.rb
@@ -0,0 +1,78 @@
+module Bosh::Director
+  module Jobs
+    class ScheduledMetricsCleanup < BaseJob
+      @queue = :normal
+
+      def self.job_type
+        :scheduled_metrics_cleanup
+      end
+
+      def self.has_work(params)
+        return false if params.first['retention_days'] <= 0
+
+        metrics_dir = Config.metrics_dir
+        return false unless File.directory?(metrics_dir)
+
+        cutoff_time = time_days_ago(params.first['retention_days'])
+
+        # Check if there are any files older than retention period
+        Dir.glob(File.join(metrics_dir, 'metric_*.bin')).any? do |file|
+          File.mtime(file) < cutoff_time
+        end
+      end
+
+      def self.time_days_ago(days)
+        Time.now - (days * 24 * 60 * 60)
+      end
+
+      def self.schedule_message
+        'clean up stale metrics files'
+      end
+
+      def initialize(params = {}) # rubocop:disable Lint/MissingSuper
+        @retention_days = params['retention_days']
+        @metrics_dir = Config.metrics_dir
+      end
+
+      def perform
+        return 'Metrics cleanup disabled (retention_days is 0)' if @retention_days <= 0
+        return "Metrics directory does not exist: #{@metrics_dir}" unless File.directory?(@metrics_dir)
+
+        cutoff_time = self.class.time_days_ago(@retention_days)
+        logger.info("Started cleanup of metrics files older than #{cutoff_time} from #{@metrics_dir}")
+
+        files_to_delete = stale_files(cutoff_time)
+        deleted_count, failed_count = delete_files(files_to_delete)
+
+        output = "Deleted #{deleted_count} metrics file(s) older than #{cutoff_time}."
+        output << " Failed to delete #{failed_count} file(s)." if failed_count.positive?
+        logger.info(output)
+        output
+      end
+
+      private
+
+      def stale_files(cutoff_time)
+        Dir.glob(File.join(@metrics_dir, 'metric_*.bin')).select do |file|
+          File.mtime(file) < cutoff_time
+        end
+      end
+
+      def delete_files(files)
+        deleted_count = 0
+        failed_count = 0
+
+        files.each do |file|
+          File.delete(file)
+          deleted_count += 1
+          logger.debug("Deleted metrics file: #{file}")
+        rescue StandardError => e
+          failed_count += 1
+          logger.warn("Failed to delete metrics file #{file}: #{e.message}")
+        end
+
+        [deleted_count, failed_count]
+      end
+    end
+  end
+end
diff --git a/src/bosh-director/spec/unit/bosh/director/jobs/scheduled_metrics_cleanup_spec.rb b/src/bosh-director/spec/unit/bosh/director/jobs/scheduled_metrics_cleanup_spec.rb
new file mode 100644
index 00000000000..6117f7b29b2
--- /dev/null
+++ b/src/bosh-director/spec/unit/bosh/director/jobs/scheduled_metrics_cleanup_spec.rb
@@ -0,0 +1,205 @@
+require 'spec_helper'
+
+module Bosh::Director
+  describe Jobs::ScheduledMetricsCleanup do
+    subject { described_class.new(*params) }
+    let(:params) do
+      [{
+        'retention_days' => retention_days,
+      }]
+    end
+    let(:retention_days) { 7 }
+    let(:metrics_dir) { Dir.mktmpdir }
+    let(:time) { Time.now }
+    let(:seven_days_seconds) { 7 * 24 * 60 * 60 }
+    let(:eight_days_ago) { time - seven_days_seconds - 86400 }
+    let(:six_days_ago) { time - seven_days_seconds + 86400 }
+
+    before do
+      allow(Config).to receive(:metrics_dir).and_return(metrics_dir)
+      allow(Time).to receive(:now).and_return(time)
+    end
+
+    after do
+      FileUtils.rm_rf(metrics_dir) if File.directory?(metrics_dir)
+    end
+
+    describe '.job_type' do
+      it 'returns the job type' do
+        expect(described_class.job_type).to eq(:scheduled_metrics_cleanup)
+      end
+    end
+
+    describe '.schedule_message' do
+      it 'outputs a message' do
+        expect(described_class.schedule_message).to eq('clean up stale metrics files')
+      end
+    end
+
+    describe '.time_days_ago' do
+      it 'calculates time correctly' do
+        expect(described_class.time_days_ago(7)).to eq(time - seven_days_seconds)
+      end
+    end
+
+    describe '.has_work' do
+      context 'when retention_days is 0' do
+        let(:retention_days) { 0 }
+
+        it 'returns false' do
+          expect(described_class.has_work(params)).to eq(false)
+        end
+      end
+
+      context 'when metrics directory does not exist' do
+        before do
+          FileUtils.rm_rf(metrics_dir)
+        end
+
+        it 'returns false' do
+          expect(described_class.has_work(params)).to eq(false)
+        end
+      end
+
+      context 'when there are stale files' do
+        before do
+          old_file = File.join(metrics_dir, 'metric_old.bin')
+          File.write(old_file, 'data')
+          File.utime(eight_days_ago, eight_days_ago, old_file)
+        end
+
+        it 'returns true' do
+          expect(described_class.has_work(params)).to eq(true)
+        end
+      end
+
+      context 'when there are no stale files' do
+        before do
+          recent_file = File.join(metrics_dir, 'metric_recent.bin')
+          File.write(recent_file, 'data')
+          File.utime(six_days_ago, six_days_ago, recent_file)
+        end
+
+        it 'returns false' do
+          expect(described_class.has_work(params)).to eq(false)
+        end
+      end
+    end
+
+    describe '#perform' do
+      context 'when retention_days is 0' do
+        let(:retention_days) { 0 }
+
+        it 'returns disabled message' do
+          expect(subject.perform).to eq('Metrics cleanup disabled (retention_days is 0)')
+        end
+      end
+
+      context 'when metrics directory does not exist' do
+        before do
+          FileUtils.rm_rf(metrics_dir)
+        end
+
+        it 'returns directory not exist message' do
+          expect(subject.perform).to eq("Metrics directory does not exist: #{metrics_dir}")
+        end
+      end
+
+      context 'when there are files to clean up' do
+        let!(:old_file_1) { File.join(metrics_dir, 'metric_old_1.bin') }
+        let!(:old_file_2) { File.join(metrics_dir, 'metric_old_2.bin') }
+        let!(:recent_file) { File.join(metrics_dir, 'metric_recent.bin') }
+        let!(:other_file) { File.join(metrics_dir, 'other_file.txt') }
+
+        before do
+          # Create old files (older than retention period)
+          File.write(old_file_1, 'data1')
+          File.utime(eight_days_ago, eight_days_ago, old_file_1)
+
+          File.write(old_file_2, 'data2')
+          File.utime(eight_days_ago, eight_days_ago, old_file_2)
+
+          # Create recent file (within retention period)
+          File.write(recent_file, 'data3')
+          File.utime(six_days_ago, six_days_ago, recent_file)
+
+          # Create non-metric file (should not be deleted)
+          File.write(other_file, 'other')
+          File.utime(eight_days_ago, eight_days_ago, other_file)
+        end
+
+        it 'deletes only old metric files' do
+          subject.perform
+
+          expect(File.exist?(old_file_1)).to eq(false)
+          expect(File.exist?(old_file_2)).to eq(false)
+          expect(File.exist?(recent_file)).to eq(true)
+          expect(File.exist?(other_file)).to eq(true)
+        end
+
+        it 'returns success message with count' do
+          cutoff_time = time - seven_days_seconds
+          expect(subject.perform).to eq("Deleted 2 metrics file(s) older than #{cutoff_time}.")
+        end
+
+        it 'logs the cleanup operation' do
+          logger = double('logger', info: nil, debug: nil, warn: nil)
+          allow(subject).to receive(:logger).and_return(logger)
+
+          subject.perform
+
+          expect(logger).to have_received(:info).at_least(:once)
+        end
+      end
+
+      context 'when file deletion fails' do
+        let!(:protected_file) { File.join(metrics_dir, 'metric_protected.bin') }
+
+        before do
+          File.write(protected_file, 'data')
+          File.utime(eight_days_ago, eight_days_ago, protected_file)
+          allow(File).to receive(:delete).with(protected_file).and_raise(Errno::EACCES, 'Permission denied')
+        end
+
+        it 'logs warning and continues' do
+          logger = double('logger', info: nil, debug: nil, warn: nil)
+          allow(subject).to receive(:logger).and_return(logger)
+
+          result = subject.perform
+
+          expect(logger).to have_received(:warn).with(/Failed to delete metrics file/)
+          expect(result).to match(/Failed to delete 1 file\(s\)/)
+        end
+
+        it 'includes failure count in result message' do
+          cutoff_time = time - seven_days_seconds
+          result = subject.perform
+          expect(result).to eq("Deleted 0 metrics file(s) older than #{cutoff_time}. Failed to delete 1 file(s).")
+        end
+      end
+
+      context 'when there are no files to clean up' do
+        it 'returns message with zero count' do
+          cutoff_time = time - seven_days_seconds
+          expect(subject.perform).to eq("Deleted 0 metrics file(s) older than #{cutoff_time}.")
+        end
+      end
+
+      context 'with different retention periods' do
+        let(:retention_days) { 30 }
+        let(:thirty_one_days_ago) { time - (31 * 24 * 60 * 60) }
+        let!(:very_old_file) { File.join(metrics_dir, 'metric_very_old.bin') }
+
+        before do
+          File.write(very_old_file, 'data')
+          File.utime(thirty_one_days_ago, thirty_one_days_ago, very_old_file)
+        end
+
+        it 'respects the configured retention period' do
+          subject.perform
+          expect(File.exist?(very_old_file)).to eq(false)
+        end
+      end
+    end
+  end
+end