diff --git a/camelot/parsers/hybrid.py b/camelot/parsers/hybrid.py
index bfa4cf38..5bf70c0e 100644
--- a/camelot/parsers/hybrid.py
+++ b/camelot/parsers/hybrid.py
@@ -247,6 +247,7 @@ def _merge_bbox_analysis(self, lattice_bbox, network_bbox):
             self.table_bbox_parses[augmented_bbox] = self.network_parser
 
     def _generate_table_bbox(self):
+        self.table_bbox_parses = {}
         # Collect bboxes from both parsers
         self.lattice_parser._generate_table_bbox()
         _lattice_bboxes = sorted(
diff --git a/tests/files/hybrid_multipage.pdf b/tests/files/hybrid_multipage.pdf
new file mode 100644
index 00000000..3d0f256d
Binary files /dev/null and b/tests/files/hybrid_multipage.pdf differ
diff --git a/tests/test_hybrid.py b/tests/test_hybrid.py
index a33f71ea..4a398737 100644
--- a/tests/test_hybrid.py
+++ b/tests/test_hybrid.py
@@ -142,3 +142,10 @@ def test_hybrid_keyerror(testdir):
     filename = os.path.join(testdir, "tabula/schools.pdf")
     tables = camelot.read_pdf(filename, flavor="hybrid", pages="4-5")
     assert len(tables) >= 1
+
+
+def test_hybrid_multipage(testdir):
+    """Hybrid parser should clear table bboxes on each new page."""
+    filename = os.path.join(testdir, "hybrid_multipage.pdf")
+    tables = camelot.read_pdf(filename, flavor="hybrid", pages="1-2")
+    assert len(tables) == 2  # not 3