diff --git a/README.md b/README.md
index 0d120d2..6900316 100644
--- a/README.md
+++ b/README.md
@@ -79,8 +79,11 @@ This design is ideal for memoization caches where:
 
 ## Limitations
 
-- **No iteration**: You cannot iterate over cached entries
-- **Eviction on collision**: Hash collisions cause immediate eviction
+- **Eviction on collision**: When two keys hash to the same bucket, the older entry is evicted.
+- **No iteration**: Individual entries cannot be enumerated.
+- **`Clone + !Drop` types**: The lock-free read path uses `!needs_drop` as a proxy for `Copy`,
+  so types that implement `Clone` with non-trivial logic but don't implement `Drop` will be
+  bitwise-copied instead of cloned on reads. Avoid using such types as keys or values.
 
 ## Feature Flags
 
diff --git a/src/lib.rs b/src/lib.rs
index 3d6f3ba..70a7283 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -106,6 +106,9 @@ impl CacheConfig for DefaultCacheConfig {}
 ///
 /// - **Eviction on collision**: When two keys hash to the same bucket, the older entry is evicted.
 /// - **No iteration**: Individual entries cannot be enumerated.
+/// - **`Clone + !Drop` types**: The lock-free read path uses `!needs_drop` as a proxy for `Copy`,
+///   so types that implement `Clone` with non-trivial logic but don't implement `Drop` will be
+///   bitwise-copied instead of cloned on reads. Avoid using such types as keys or values.
 ///
 /// # Type Parameters
 ///
@@ -978,6 +981,11 @@ mod tests {
 
     #[test]
     fn concurrent_read_write() {
+        // Miri flags the seqlock's speculative read as a data race with concurrent writers.
+        if cfg!(miri) {
+            return;
+        }
+
         let cache: Cache<u64, u64> = new_cache(256);
         let n = iters(1000);
 
@@ -994,6 +1002,7 @@ mod tests {
 
     #[test]
     fn seqlock_aba() {
+        // Miri flags the seqlock's speculative read as a data race with concurrent writers.
         if cfg!(miri) {
             return;
         }
@@ -1020,6 +1029,11 @@ mod tests {
 
     #[test]
     fn concurrent_get_or_insert() {
+        // Miri flags the seqlock's speculative read as a data race with concurrent writers.
+        if cfg!(miri) {
+            return;
+        }
+
         let cache: Cache<u64, u64> = new_cache(1024);
         let n = iters(100);
 
@@ -1368,6 +1382,7 @@ mod tests {
 
     #[test]
     fn epoch_concurrent_seqlock() {
+        // Miri flags the seqlock's speculative read as a data race with concurrent writers.
         if cfg!(miri) {
             return;
         }
diff --git a/src/stats.rs b/src/stats.rs
index 1cc781d..4f5563a 100644
--- a/src/stats.rs
+++ b/src/stats.rs
@@ -513,4 +513,42 @@ mod tests {
         cache.remove(&CollidingKey(99, 99));
         assert_eq!(handler.removes(), 1);
     }
+
+    #[test]
+    fn concurrent_stats() {
+        // Miri flags the seqlock's speculative read as a data race with concurrent writers.
+        if cfg!(miri) {
+            return;
+        }
+
+        let handler = Arc::new(CountingStatsHandler::new());
+        let stats = Stats::new(Arc::clone(&handler));
+        let cache: Cache<u64, u64, BH> =
+            Cache::new(1024, Default::default()).with_stats(Some(stats));
+
+        std::thread::scope(|s| {
+            for t in 0..4 {
+                let cache = &cache;
+                s.spawn(move || {
+                    for i in 0..1000u64 {
+                        match t {
+                            0 => cache.insert(i % 100, i),
+                            1 => _ = cache.get(&(i % 100)),
+                            2 => _ = cache.get_or_insert_with(i % 100, |&k| k * 2),
+                            _ => _ = cache.remove(&(i % 100)),
+                        }
+                    }
+                });
+            }
+        });
+
+        let total = handler.hits()
+            + handler.misses()
+            + handler.inserts()
+            + handler.updates()
+            + handler.removes()
+            + handler.collisions();
+        assert!(total > 0);
+        assert_eq!(handler.hits() + handler.misses(), 1000 + 1000);
+    }
 }