diff --git a/test/mail_with_postgrey.bats b/test/mail_with_postgrey.bats
index 71149a07..03248295 100644
--- a/test/mail_with_postgrey.bats
+++ b/test/mail_with_postgrey.bats
@@ -1,15 +1,11 @@
 load 'test_helper/common'
 
 function setup() {
-    if [ "$BATS_TEST_NUMBER" -eq 1 ]; then
-        setup_file
-    fi
+    run_setup_file_if_necessary
 }
 
 function teardown() {
-    if [ "$BATS_TEST_NUMBER" -eq ${#BATS_TEST_NAMES[@]} ]; then
-        teardown_file
-    fi
+    run_teardown_file_if_necessary
 }
 
 function setup_file() {
@@ -31,6 +27,10 @@ function teardown_file() {
     docker rm -f mail_with_postgrey
 }
 
+@test "first" {
+  # this test must come first to reliably identify when to run setup_file
+}
+
 @test "checking postgrey: /etc/postfix/main.cf correctly edited" {
   run docker exec mail_with_postgrey /bin/bash -c "grep 'bl.spamcop.net, check_policy_service inet:127.0.0.1:10023' /etc/postfix/main.cf | wc -l"
   assert_success
@@ -94,4 +94,8 @@ function teardown_file() {
   run docker exec mail_with_postgrey /bin/sh -c "grep -i 'action=pass, reason=recipient whitelist' /var/log/mail/mail.log | wc -l"
   assert_success
   assert_output 1
+}
+
+@test "last" {
+  # this test is only there to reliably mark the end for the teardown_file
 }
\ No newline at end of file
diff --git a/test/mail_with_relays.bats b/test/mail_with_relays.bats
index 58447859..fa2ea0ef 100644
--- a/test/mail_with_relays.bats
+++ b/test/mail_with_relays.bats
@@ -1,15 +1,11 @@
 load 'test_helper/common'
 
 function setup() {
-    if [ "$BATS_TEST_NUMBER" -eq 1 ]; then
-        setup_file
-    fi
+    run_setup_file_if_necessary
 }
 
 function teardown() {
-    if [ "$BATS_TEST_NUMBER" -eq ${#BATS_TEST_NAMES[@]} ]; then
-        teardown_file
-    fi
+    run_teardown_file_if_necessary
 }
 
 function setup_file() {
@@ -31,6 +27,10 @@ function teardown_file() {
     docker rm -f mail_with_relays
 }
 
+@test "first" {
+  # this test must come first to reliably identify when to run setup_file
+}
+
 @test "checking relay hosts: default mapping is added from env vars" {
   run docker exec mail_with_relays /bin/sh -c 'cat /etc/postfix/relayhost_map | grep -e "^@domainone.tld\s\+\[default.relay.com\]:2525" | wc -l | grep 1'
   assert_success
@@ -54,4 +54,8 @@ function teardown_file() {
 @test "checking relay hosts: default auth entry is added" {
   run docker exec mail_with_relays /bin/sh -c 'cat /etc/postfix/sasl_passwd | grep -e "^\[default.relay.com\]:2525\s\+smtp_user:smtp_password" | wc -l | grep 1'
   assert_success
+}
+
+@test "last" {
+  # this test is only there to reliably mark the end for the teardown_file
 }
\ No newline at end of file
diff --git a/test/test_helper/common.bash b/test/test_helper/common.bash
index 3ee7c7d4..5e119f99 100644
--- a/test/test_helper/common.bash
+++ b/test/test_helper/common.bash
@@ -4,7 +4,7 @@ load 'test_helper/bats-assert/load'
 NAME=tvial/docker-mailserver:testing
 
 # default timeout is 60 seconds
-TEST_TIMEOUT_IN_SECONDS=${TIMEOUT-60}
+TEST_TIMEOUT_IN_SECONDS=${TEST_TIMEOUT_IN_SECONDS-60}
 
 function repeat_until_success_or_timeout {
     if ![[ "$1" ~= '^[0-9]+$' ]]; then
@@ -32,4 +32,18 @@ function wait_for_smtp_port_in_container() {
 # @param $1 name of the postfix container
 function wait_for_finished_setup_in_container() {
     repeat_until_success_or_timeout $TEST_TIMEOUT_IN_SECONDS sh -c "docker logs $1 | grep 'Starting mail server'"
+}
+
+# use in setup() in conjunction with a `@test "first" {}` to trigger setup_file reliably
+function run_setup_file_if_necessary() {
+    if [ "$BATS_TEST_NAME" == 'test_first' ]; then
+        setup_file
+    fi
+}
+
+# use in teardown() in conjunction with a `@test "last" {}` to trigger teardown_file reliably
+function run_teardown_file_if_necessary() {
+    if [ "$BATS_TEST_NAME" == 'test_last' ]; then
+        teardown_file
+    fi
 }
\ No newline at end of file