diff --git a/include/verilated_vpi.cpp b/include/verilated_vpi.cpp
index 0ca223349..6554acba2 100644
--- a/include/verilated_vpi.cpp
+++ b/include/verilated_vpi.cpp
@@ -2779,7 +2779,8 @@ VarAccessInfo<T> vl_vpi_var_access_info(const VerilatedVpioVarBase* vop, size_t
 
     // make sure we're not trying to write outside var bounds
     assert(varBits > addOffset);
-    bitCount = std::min(bitCount, varBits - addOffset);
+    bitCount = std::min({wordBits,  // For VLWide, varBits can exceed wordBits
+                         bitCount, varBits - addOffset});
 
     VarAccessInfo<T> info;
     info.m_datap = reinterpret_cast<T*>(vop->varDatap());
diff --git a/test_regress/t/t_vpi_var.cpp b/test_regress/t/t_vpi_var.cpp
index 9c1686c2d..364968244 100644
--- a/test_regress/t/t_vpi_var.cpp
+++ b/test_regress/t/t_vpi_var.cpp
@@ -1288,11 +1288,17 @@ int _mon_check_multi_index() {
         CHECK_RESULT_Z(vh_oob);
     }
 
-    // Multiple packed dimensions: multi_packed is [3:0][7:0] multi_packed[2:0]
+    // Multiple packed dimensions: multi_packed is [15:0][7:0] multi_packed[2:0]
     {
         TestVpiHandle vh1 = vpi_handle_by_name((PLI_BYTE8*)"t.multi_packed[1]", nullptr);
         CHECK_RESULT_NZ(vh1);
-        CHECK_RESULT(vpi_get(vpiSize, vh1), 32);  // 4*8
+        CHECK_RESULT(vpi_get(vpiSize, vh1), 128);  // 16*8
+
+        // Write last 32 bits of the packed vector in the specified unpacked dimension,
+        // i.e. the four 8-bit elements in multi_packed[1][3:0]
+        v.value.integer = 0xAABBCCDD;
+        vpi_put_value(vh1, &v, nullptr, vpiNoDelay);
+
         // Index into first packed dim -> 8-bit sub-word
         TestVpiHandle vh2 = vpi_handle_by_index(vh1, 2);
         CHECK_RESULT_NZ(vh2);
@@ -1301,6 +1307,12 @@ int _mon_check_multi_index() {
         TestVpiHandle vh3 = vpi_handle_by_index(vh2, 3);
         CHECK_RESULT_NZ(vh3);
         CHECK_RESULT(vpi_get(vpiSize, vh3), 1);
+
+        // Index into the last bits of the packed array and check value
+        TestVpiHandle vh_last = vpi_handle_by_index(vh1, 0);
+        CHECK_RESULT_NZ(vh_last);
+        vpi_get_value(vh_last, &v);
+        CHECK_RESULT(v.value.integer, 0xDD);
     }
 
     // Partial indexing (not all unpacked dimensions)
diff --git a/test_regress/t/t_vpi_var.v b/test_regress/t/t_vpi_var.v
index 553371661..df59e4e89 100644
--- a/test_regress/t/t_vpi_var.v
+++ b/test_regress/t/t_vpi_var.v
@@ -50,7 +50,7 @@ extern "C" int mon_check();
   reg [0:95]      mem_3d[0:1][1:0][0:1]  /*verilator public_flat_rw */;  // Mixed: asc, desc, asc
   // verilator lint_on ASCRANGE
 
-  reg [3:0] [7:0] multi_packed[2:0]  /*verilator public_flat_rw */;
+  reg [15:0] [7:0] multi_packed[2:0]  /*verilator public_flat_rw */;
   reg             unpacked_only[7:0];
 
   reg [7:0]       text_byte    /*verilator public_flat_rw @(posedge clk) */;
@@ -124,8 +124,8 @@ extern "C" int mon_check();
     end
 
     for (int i = 0; i < 3; i++) begin
-      for (int j = 0; j < 4; j++) begin
-        multi_packed[i][j] = 8'((i * 4) + j);
+      for (int j = 0; j < 16; j++) begin
+        multi_packed[i][j] = 8'((i * 16) + j);
       end
     end
 
diff --git a/test_regress/t/t_vpi_var2.v b/test_regress/t/t_vpi_var2.v
index 5d1d0136d..14dd8f7f2 100644
--- a/test_regress/t/t_vpi_var2.v
+++ b/test_regress/t/t_vpi_var2.v
@@ -66,7 +66,7 @@ extern "C" int mon_check();
   // verilator lint_on ASCRANGE
 
   // Signal with multiple packed dimensions
-  reg [3:0] [7:0] multi_packed[2:0];
+  reg [15:0] [7:0] multi_packed[2:0];
   reg             unpacked_only[7:0];
 /*verilator public_off*/
   reg             invisible2;
@@ -142,8 +142,8 @@ extern "C" int mon_check();
     end
 
     for (int i = 0; i < 3; i++) begin
-      for (int j = 0; j < 4; j++) begin
-        multi_packed[i][j] = 8'((i * 4) + j);
+      for (int j = 0; j < 16; j++) begin
+        multi_packed[i][j] = 8'((i * 16) + j);
       end
     end
 
diff --git a/test_regress/t/t_vpi_var3.v b/test_regress/t/t_vpi_var3.v
index 930c6c14e..e87bc8f68 100644
--- a/test_regress/t/t_vpi_var3.v
+++ b/test_regress/t/t_vpi_var3.v
@@ -51,7 +51,7 @@ extern "C" int mon_check();
   // verilator lint_on ASCRANGE
 
   // Signal with multiple packed dimensions
-  reg [3:0] [7:0] multi_packed[2:0];
+  reg [15:0] [7:0] multi_packed[2:0];
   reg             unpacked_only[7:0];
 
   reg [7:0]       text_byte;
@@ -122,8 +122,8 @@ extern "C" int mon_check();
     end
 
     for (int i = 0; i < 3; i++) begin
-      for (int j = 0; j < 4; j++) begin
-        multi_packed[i][j] = 8'((i * 4) + j);
+      for (int j = 0; j < 16; j++) begin
+        multi_packed[i][j] = 8'((i * 16) + j);
       end
     end