diff --git a/tests/gold_tests/logging/log-milestone-fields.test.py b/tests/gold_tests/logging/log-milestone-fields.test.py
index 60fc69f4536..e2a47ccb346 100644
--- a/tests/gold_tests/logging/log-milestone-fields.test.py
+++ b/tests/gold_tests/logging/log-milestone-fields.test.py
@@ -33,7 +33,7 @@ class MilestoneFieldsTest:
     cache result code.  A validation script then parses the log and checks:
 
     - Every expected key=value pair is present on each line
-    - All values are integers (>= 0 or -1 for unset milestones)
+    - All values are integers, with "-" for unset milestones
     - Cache miss line: ms > 0, origin-phase fields present
     - Cache hit line: hit_proc >= 0 and hit_xfer >= 0
     - No epoch-length garbage values (> 1_000_000_000)
@@ -155,9 +155,12 @@ def _sendCacheHit(self):
         tr.StillRunningAfter = self._ts
 
     def _waitForLog(self):
-        tr = Test.AddTestRun('Wait for log file to be written')
-        tr.Processes.Default.Command = (os.path.join(Test.Variables.AtsTestToolsDir, 'condwait') + f' 60 1 -f {self._log_path}')
-        tr.Processes.Default.ReturnCode = 0
+        tr = Test.AddAwaitFileContainsTestRun(
+            'Wait for milestone log lines to be written',
+            self._log_path,
+            r'^crc=.* hit_xfer=',
+            desired_count=2,
+        )
         tr.StillRunningAfter = self._server
         tr.StillRunningAfter = self._ts
 
diff --git a/tests/gold_tests/logging/verify_milestone_fields.py b/tests/gold_tests/logging/verify_milestone_fields.py
index dd161cfccdd..419b8ccc9a3 100644
--- a/tests/gold_tests/logging/verify_milestone_fields.py
+++ b/tests/gold_tests/logging/verify_milestone_fields.py
@@ -54,6 +54,9 @@
 MISS_CHAIN = ['c_hdr', 'c_proc', 'cache', 'dns', 'o_tcp', 'o_wait', 'o_hdr', 'o_proc']
 
 EPOCH_THRESHOLD = 1_000_000_000
+# Each msdms field is truncated independently to integer milliseconds.
+# Allow one truncated millisecond per component in the miss chain.
+CHAIN_TOLERANCE = len(MISS_CHAIN)
 
 
 def parse_line(line: str) -> dict[str, str]:
@@ -116,7 +119,8 @@ def validate_line(fields: dict[str, str], line_num: int) -> list[str]:
             if val < -10:
                 errors.append(f'line {line_num}: miss field "{name}" has unexpected value: {val}')
 
-        # Verify chain sum approximates c_ttfb (within tolerance for rounding).
+        # Verify the signed chain sum approximates c_ttfb within a tolerance
+        # for per-field millisecond truncation.
         chain_vals = []
         for name in MISS_CHAIN:
             val_str = fields.get(name)
@@ -124,8 +128,7 @@ def validate_line(fields: dict[str, str], line_num: int) -> list[str]:
                 chain_vals.append(0)
                 continue
             try:
-                v = int(val_str)
-                chain_vals.append(v if v >= 0 else 0)
+                chain_vals.append(int(val_str))
             except ValueError:
                 chain_vals.append(0)
 
@@ -134,8 +137,7 @@ def validate_line(fields: dict[str, str], line_num: int) -> list[str]:
         if c_ttfb_str and c_ttfb_str != '-':
             try:
                 c_ttfb_val = int(c_ttfb_str)
-                # Allow 2ms tolerance for rounding across multiple sub-millisecond fields.
-                if c_ttfb_val >= 0 and abs(chain_sum - c_ttfb_val) > 2:
+                if c_ttfb_val >= 0 and abs(chain_sum - c_ttfb_val) > CHAIN_TOLERANCE:
                     errors.append(
                         f'line {line_num}: chain sum ({chain_sum}) != c_ttfb ({c_ttfb_val}), '
                         f'diff={abs(chain_sum - c_ttfb_val)}ms')