diff --git a/fragment.go b/fragment.go
index b4dc828d1..149749230 100644
--- a/fragment.go
+++ b/fragment.go
@@ -2249,7 +2249,7 @@ func (f *fragment) Snapshot() error {
 func track(start time.Time, message string, stats stats.StatsClient, logger logger.Logger) {
 	elapsed := time.Since(start)
 	logger.Debugf("%s took %s", message, elapsed)
-	stats.Histogram(MetricSnapshotDurationSeconds, elapsed.Seconds(), 1.0)
+	stats.Timing(MetricSnapshotDurationSeconds, elapsed, 1.0)
 }
 
 // snapshot does the actual snapshot operation. it does not check or care
diff --git a/holder.go b/holder.go
index 247bce041..51a89826b 100644
--- a/holder.go
+++ b/holder.go
@@ -801,10 +801,10 @@ func (s *holderSyncer) SyncHolder() error {
 					}
 				}
 			}
-			s.Stats.Histogram(MetricSyncFieldDurationSeconds, float64(time.Since(tf)), 1.0)
+			s.Stats.Timing(MetricSyncFieldDurationSeconds, time.Since(tf), 1.0)
 			tf = time.Now() // reset tf
 		}
-		s.Stats.Histogram(MetricSyncIndexDurationSeconds, float64(time.Since(ti)), 1.0)
+		s.Stats.Timing(MetricSyncIndexDurationSeconds, time.Since(ti), 1.0)
 		ti = time.Now() // reset ti
 	}
 
diff --git a/server.go b/server.go
index 957f45100..ab90bbc9a 100644
--- a/server.go
+++ b/server.go
@@ -675,7 +675,7 @@ func (s *Server) monitorAntiEntropy() {
 		// Record successful sync in log.
 		s.logger.Printf("holder sync complete")
 		dif := time.Since(t)
-		s.holder.Stats.Histogram(MetricAntiEntropyDurationSeconds, float64(dif), 1.0)
+		s.holder.Stats.Timing(MetricAntiEntropyDurationSeconds, dif, 1.0)
 
 		// Drain tick channel since we just finished anti-entropy. If the AE
 		// process took a long time, we don't want them to pile up on each