diff --git a/src/metadata_writer.rs b/src/metadata_writer.rs
index 15ca7b6..3339262 100644
--- a/src/metadata_writer.rs
+++ b/src/metadata_writer.rs
@@ -106,7 +106,18 @@ pub struct DataFileInfo {
 
 impl DataFileInfo {
     /// Create a new data file info with relative path.
+    ///
+    /// # Panics
+    ///
+    /// Panics if `record_count` is negative. Record counts originate from
+    /// `RecordBatch::num_rows()` (always non-negative), so a negative value
+    /// indicates a programming error.
     pub fn new(path: impl Into<String>, file_size_bytes: i64, record_count: i64) -> Self {
+        assert!(
+            record_count >= 0,
+            "record_count must be non-negative, got {}",
+            record_count
+        );
         Self {
             path: path.into(),
             path_is_relative: true,
@@ -300,4 +311,16 @@ mod tests {
         let file = DataFileInfo::new("/absolute/path.parquet", 1024, 100).with_absolute_path();
         assert!(!file.path_is_relative);
     }
+
+    #[test]
+    fn test_data_file_info_zero_record_count() {
+        let file = DataFileInfo::new("empty.parquet", 0, 0);
+        assert_eq!(file.record_count, 0);
+    }
+
+    #[test]
+    #[should_panic(expected = "record_count must be non-negative")]
+    fn test_data_file_info_negative_record_count_panics() {
+        DataFileInfo::new("test.parquet", 1024, -1);
+    }
 }
diff --git a/src/table.rs b/src/table.rs
index ed8870a..ded6d5c 100644
--- a/src/table.rs
+++ b/src/table.rs
@@ -63,6 +63,20 @@ pub(crate) fn validated_file_size(file_size_bytes: i64, file_path: &str) -> Data
     })
 }
 
+/// Validate and convert record_count from i64 (as stored in DuckLake metadata) to u64.
+///
+/// DuckLake stores record counts as signed integers in SQL. A negative value indicates
+/// corrupt or invalid metadata. Without this check, a negative record_count would cause
+/// incorrect behavior (e.g., empty ranges in full-file deletes, or incorrect row filtering).
+pub(crate) fn validated_record_count(record_count: i64, file_path: &str) -> DataFusionResult<u64> {
+    u64::try_from(record_count).map_err(|_| {
+        DataFusionError::Execution(format!(
+            "Invalid record_count ({}) for file '{}': value must be non-negative",
+            record_count, file_path
+        ))
+    })
+}
+
 /// Returns the expected schema for DuckLake delete files
 ///
 /// Delete files have a standard schema: (file_path: VARCHAR, pos: INT64)
@@ -731,4 +745,43 @@ mod tests {
         assert!(msg.contains("bad.parquet"));
         assert!(msg.contains(&i64::MIN.to_string()));
     }
+
+    #[test]
+    fn test_validated_record_count_positive() {
+        assert_eq!(validated_record_count(0, "test.parquet").unwrap(), 0);
+        assert_eq!(validated_record_count(100, "test.parquet").unwrap(), 100);
+        assert_eq!(
+            validated_record_count(i64::MAX, "test.parquet").unwrap(),
+            i64::MAX as u64
+        );
+    }
+
+    #[test]
+    fn test_validated_record_count_negative() {
+        let err = validated_record_count(-1, "data/test.parquet").unwrap_err();
+        let msg = err.to_string();
+        assert!(
+            msg.contains("-1"),
+            "Error should contain the negative value: {}",
+            msg
+        );
+        assert!(
+            msg.contains("data/test.parquet"),
+            "Error should contain the file path: {}",
+            msg
+        );
+        assert!(
+            msg.contains("record_count"),
+            "Error should mention record_count: {}",
+            msg
+        );
+    }
+
+    #[test]
+    fn test_validated_record_count_large_negative() {
+        let err = validated_record_count(i64::MIN, "bad.parquet").unwrap_err();
+        let msg = err.to_string();
+        assert!(msg.contains("bad.parquet"));
+        assert!(msg.contains(&i64::MIN.to_string()));
+    }
 }
diff --git a/src/table_deletions.rs b/src/table_deletions.rs
index 9c3b6a4..8ad73c7 100644
--- a/src/table_deletions.rs
+++ b/src/table_deletions.rs
@@ -36,7 +36,7 @@ use futures::Stream;
 
 use crate::metadata_provider::{DeleteFileChange, MetadataProvider};
 use crate::path_resolver::resolve_path;
-use crate::table::validated_file_size;
+use crate::table::{validated_file_size, validated_record_count};
 
 /// Delete file schema: (file_path: VARCHAR, pos: INT64)
 fn delete_file_schema() -> SchemaRef {
@@ -143,6 +143,10 @@ impl TableDeletionsTable {
             delete_file.data_file_footer_size,
         )?;
 
+        // Validate record_count before use — a negative value from corrupt metadata
+        // would cause incorrect behavior (e.g., empty ranges in full-file deletes).
+        validated_record_count(delete_file.data_record_count, &delete_file.data_file_path)?;
+
         Ok(Arc::new(DeletedRowsExec::new(
             current_delete_exec,
             previous_delete_exec,
diff --git a/tests/common/mod.rs b/tests/common/mod.rs
index 8d4b770..b6081c5 100644
--- a/tests/common/mod.rs
+++ b/tests/common/mod.rs
@@ -14,12 +14,17 @@ use std::sync::Once;
 
 static INSTALL_DUCKLAKE: Once = Once::new();
 
-/// Install the ducklake extension exactly once (thread-safe across parallel tests).
+/// Install the ducklake and parquet extensions exactly once (thread-safe across parallel tests).
+///
+/// Pre-installing parquet avoids race conditions when multiple concurrent tests
+/// trigger DuckDB's auto-install simultaneously (observed on macOS CI).
 fn ensure_ducklake_installed() {
     INSTALL_DUCKLAKE.call_once(|| {
         let conn = duckdb::Connection::open_in_memory().expect("open in-memory duckdb");
         conn.execute("INSTALL ducklake;", [])
             .expect("install ducklake extension");
+        conn.execute("INSTALL parquet;", [])
+            .expect("install parquet extension");
     });
 }