diff --git a/wolfcrypt/src/sp_int.c b/wolfcrypt/src/sp_int.c
index 99287dfee..9c05fb883 100644
--- a/wolfcrypt/src/sp_int.c
+++ b/wolfcrypt/src/sp_int.c
@@ -11896,10 +11896,12 @@ static int _sp_mont_red(sp_int* a, sp_int* m, sp_int_digit mp)
     else if (m->used == 4) {
         sp_int_digit l;
         sp_int_digit h;
+        sp_int_digit o2;
 
         l = 0;
         h = 0;
         o = 0;
+        o2 = 0;
         for (i = 0; i < 4; i++) {
             mu = mp * a->dp[i];
             if ((i == 3) && (mask != 0)) {
@@ -11919,15 +11921,17 @@ static int _sp_mont_red(sp_int* a, sp_int* m, sp_int_digit mp)
             SP_ASM_MUL_ADD_NO(l, h, mu, m->dp[2]);
             a->dp[i + 2] = l;
             l = h;
-            h = 0;
+            h = o2;
+            o2 = 0;
             SP_ASM_ADDC_REG(l, h, o);
             SP_ASM_ADDC(l, h, a->dp[i + 3]);
-            SP_ASM_MUL_ADD_NO(l, h, mu, m->dp[3]);
+            SP_ASM_MUL_ADD(l, h, o2, mu, m->dp[3]);
             a->dp[i + 3] = l;
             o = h;
             l = h;
             h = 0;
         }
+        h = o2;
         SP_ASM_ADDC(l, h, a->dp[7]);
         a->dp[7] = l;
         a->dp[8] = h;
@@ -11936,10 +11940,12 @@ static int _sp_mont_red(sp_int* a, sp_int* m, sp_int_digit mp)
     else if (m->used == 6) {
         sp_int_digit l;
         sp_int_digit h;
+        sp_int_digit o2;
 
         l = 0;
         h = 0;
         o = 0;
+        o2 = 0;
         for (i = 0; i < 6; i++) {
             mu = mp * a->dp[i];
             if ((i == 5) && (mask != 0)) {
@@ -11969,15 +11975,17 @@ static int _sp_mont_red(sp_int* a, sp_int* m, sp_int_digit mp)
             SP_ASM_MUL_ADD_NO(l, h, mu, m->dp[4]);
             a->dp[i + 4] = l;
             l = h;
-            h = 0;
+            h = o2;
+            o2 = 0;
             SP_ASM_ADDC_REG(l, h, o);
             SP_ASM_ADDC(l, h, a->dp[i + 5]);
-            SP_ASM_MUL_ADD_NO(l, h, mu, m->dp[5]);
+            SP_ASM_MUL_ADD(l, h, o2, mu, m->dp[5]);
             a->dp[i + 5] = l;
             o = h;
             l = h;
             h = 0;
         }
+        h = o2;
         SP_ASM_ADDC(l, h, a->dp[11]);
         a->dp[11] = l;
         a->dp[12] = h;
diff --git a/wolfcrypt/test/test.c b/wolfcrypt/test/test.c
index 7cd327650..84d6104e5 100644
--- a/wolfcrypt/test/test.c
+++ b/wolfcrypt/test/test.c
@@ -35851,6 +35851,7 @@ static int mp_test_mont(mp_int* a, mp_int* m, mp_int* n, mp_int* r, WC_RNG* rng)
                                0x01,  0x9f,  0x13,  0xbd,
                                0x1f, 0x13d,  0x45, 0x615
                             };
+    int bits[] = { 256, 384, 2048, 3072 };
     int i;
     int j;
 
@@ -35908,6 +35909,31 @@ static int mp_test_mont(mp_int* a, mp_int* m, mp_int* n, mp_int* r, WC_RNG* rng)
         }
     }
 
+    /* Force carries. */
+    for (i = 0; i < (int)(sizeof(bits) / sizeof(*bits)); i++) {
+        /* a = 2^(bits*2) - 1 */
+        mp_zero(a);
+        mp_set_bit(a, bits[i] * 2);
+        mp_sub_d(a, 1, a);
+        /* m = 2^(bits) - 1 */
+        mp_zero(m);
+        mp_set_bit(m, bits[i]);
+        mp_sub_d(m, 1, m);
+        mp = 1;
+        /* result = r = 2^(bits) - 1 */
+        mp_zero(r);
+        mp_set_bit(r, bits[i]);
+        mp_sub_d(r, 1, r);
+
+        ret = mp_montgomery_reduce(a, m, mp);
+        if (ret != MP_OKAY)
+            return -13240;
+
+        /* Result is m or 0 if reduced to range of modulus. */
+        if (mp_cmp(a, r) != MP_EQ && mp_iszero(a) != MP_YES)
+            return -13241;
+    }
+
     return 0;
 }
 #endif