diff --git a/cardano_node_tests/utils/logfiles.py b/cardano_node_tests/utils/logfiles.py
index 3ba3c9a39..dad4224de 100644
--- a/cardano_node_tests/utils/logfiles.py
+++ b/cardano_node_tests/utils/logfiles.py
@@ -512,7 +512,7 @@ def _validated_start(seek: int | None, size: int) -> int:
 def _search_log_lines(  # noqa: C901
     logfile: pl.Path,
     rotated_logs: list[RotableLog],
-    errors_re: re.Pattern[str],  # The the error regex needs to be unanchored
+    errors_re: re.Pattern[str],  # The error regex needs to be unanchored
     *,
     errors_ignored_re: re.Pattern[str] | None = None,
     look_back_map: dict[str, str] | None = None,
@@ -1016,10 +1016,10 @@ def _search(
             # and inode of the log file the seek offset was recorded for
             seek, timestamp, inode = _load_search_state(logfile=logfile)
 
-            # Get ignore rules for the log file
-            ignore_rules = _get_ignore_rules(
-                cluster_env=cluster_env, timestamp=timestamp or time.time()
-            )
+            # Get ignore rules for the log file. A log file that was not searched yet (or
+            # whose offset file was lost) has timestamp 0.0, so no rule is expired for it -
+            # the whole file is going to be searched and the rules apply to all of it.
+            ignore_rules = _get_ignore_rules(cluster_env=cluster_env, timestamp=timestamp)
             errors_ignored = _get_ignore_regex(
                 ignore_rules=ignore_rules, regexes=ERRORS_IGNORED, logfile=logfile
             )
diff --git a/framework_tests/test_logfiles.py b/framework_tests/test_logfiles.py
index f7d0463d0..8f5993dd2 100644
--- a/framework_tests/test_logfiles.py
+++ b/framework_tests/test_logfiles.py
@@ -932,3 +932,29 @@ def test_constrains_match_end(pattern: str, expected: bool):
     """
     regex_b = re.compile(pattern.encode("utf-8"))
     assert logfiles._constrains_match_end(regex_b) is expected
+
+
+def test_search_cluster_logs_first_search_expiry(cluster_env: cluster_nodes.ClusterEnv):
+    """Check that ignore rules apply to the first search of a log file.
+
+    A log file that was not searched yet may contain ignored errors from any time in
+    the past. An ignore rule must not expire for such file, even when its expire time
+    already passed. For further searches the rule is expired.
+    """
+    logfile = _write_log(
+        state_dir=cluster_env.state_dir, name="node1.stdout", content="ignored error one\n"
+    )
+    # The expire time is far in the past
+    logfiles.add_ignore_rule(
+        files_glob="*.stdout", regex="ignored error", ignore_file_id="id1", skip_after=100.0
+    )
+
+    # First search: the rule applies to the whole unsearched file history
+    assert logfiles.search_cluster_logs() == []
+
+    # Further searches: the rule is expired, new occurrences are reported
+    with open(logfile, "a", encoding="utf-8") as outfile:
+        outfile.write("ignored error two\n")
+
+    errors = logfiles.search_cluster_logs()
+    assert [e[1] for e in errors] == ["ignored error two"]