diff --git a/tests/unit-mcdc/test_dh_fault_whitebox.c b/tests/unit-mcdc/test_dh_fault_whitebox.c
index 1f5ca11903d..d1c57037276 100644
--- a/tests/unit-mcdc/test_dh_fault_whitebox.c
+++ b/tests/unit-mcdc/test_dh_fault_whitebox.c
@@ -39,7 +39,7 @@
  *   (b) size-capacity rows: sp_int is a fixed SP_INT_DIGITS-digit struct
  *       (SP_INT_BITS=4096 here); mp_read_unsigned_bin()/sp_exptmod() reject
  *       inputs that do not fit *by construction* (sp_read_unsigned_bin: inSz
- *       > a->size*SP_WORD_SIZEOF; sp_exptmod_ex: m->used*2 >= SP_INT_DIGITS).
+ *       > a->size*SP_WORD_OCTETS; sp_exptmod_ex: m->used*2 >= SP_INT_DIGITS).
  *       Passing an oversized buffer (here: 2000 zero bytes, always over the
  *       ~1024-1032 byte capacity regardless of 32/64-bit sp_int digits) or a
  *       modulus bigger than SP_INT_BITS deterministically fails that call
diff --git a/wolfcrypt/src/sp_int.c b/wolfcrypt/src/sp_int.c
index 24d65221cad..0296eb44470 100644
--- a/wolfcrypt/src/sp_int.c
+++ b/wolfcrypt/src/sp_int.c
@@ -5426,7 +5426,7 @@ void sp_forcezero(sp_int* a)
     /* Zeroize when a valid pointer passed in. */
     if (a != NULL) {
         /* Ensure all data zeroized - data not zeroed when used decreases. */
-        ForceZero(a->dp, a->size * (word32)SP_WORD_SIZEOF);
+        ForceZero(a->dp, a->size * (word32)sizeof(sp_int_digit));
         /* Set back to zero. */
     #ifdef HAVE_WOLF_BIGINT
         /* Zeroize the raw data as well. */
@@ -5453,7 +5453,7 @@ static void _sp_copy(const sp_int* a, sp_int* r)
         r->dp[0] = 0;
     }
     else {
-        XMEMCPY(r->dp, a->dp, a->used * (word32)SP_WORD_SIZEOF);
+        XMEMCPY(r->dp, a->dp, a->used * (word32)sizeof(sp_int_digit));
     }
     /* Set number of used words in result. */
     r->used = a->used;/* // NOLINT(clang-analyzer-core.uninitialized.Assign) */
@@ -6217,7 +6217,7 @@ int sp_set_bit(sp_int* a, int i)
      * check below rather than wrapping. */
     int wi = (i < 0) ? 0 : (i >> SP_WORD_SHIFT);
 
-#if SP_INT_DIGITS < (65536 / SP_WORD_SIZEOF)
+#if SP_INT_DIGITS < (65536 / SP_WORD_OCTETS)
     /* Check bit index isn't bigger than maximum allowed. */
     if (i > SP_INT_DIGITS * SP_WORD_SIZE) {
         err = MP_VAL;
@@ -8584,9 +8584,9 @@ int sp_lshd(sp_int* a, int s)
     }
     if (err == MP_OKAY) {
         /* Move up digits. */
-        XMEMMOVE(a->dp + s, a->dp, a->used * (word32)SP_WORD_SIZEOF);
+        XMEMMOVE(a->dp + s, a->dp, a->used * (word32)sizeof(sp_int_digit));
         /* Back fill with zeros. */
-        XMEMSET(a->dp, 0, (size_t)s * SP_WORD_SIZEOF);
+        XMEMSET(a->dp, 0, (size_t)s * sizeof(sp_int_digit));
         /* Update used. */
         a->used = (sp_size_t)(a->used + s);
         /* Remove leading zeros. */
@@ -8652,13 +8652,14 @@ static int sp_lshb(sp_int* a, int n)
             /* Only digits to move and ensure not zero. */
             else if (s > 0) {
                 /* Move up digits. */
-                XMEMMOVE(a->dp + s, a->dp, a->used * (word32)SP_WORD_SIZEOF);
+                XMEMMOVE(a->dp + s, a->dp,
+                    a->used * (word32)sizeof(sp_int_digit));
             }
 
             /* Update used digit count. */
             a->used = (sp_size_t)(a->used + s);
             /* Back fill with zeros. */
-            XMEMSET(a->dp, 0, (word32)SP_WORD_SIZEOF * s);
+            XMEMSET(a->dp, 0, (word32)sizeof(sp_int_digit) * s);
         }
     }
 
@@ -8735,10 +8736,12 @@ int sp_rshb(const sp_int* a, int n, sp_int* r)
             r->used = (sp_size_t)(a->used - i);
             /* Move digits down. */
             if (r == a) {
-                XMEMMOVE(r->dp, r->dp + i, (word32)SP_WORD_SIZEOF * r->used);
+                XMEMMOVE(r->dp, r->dp + i,
+                    (word32)sizeof(sp_int_digit) * r->used);
             }
             else {
-                XMEMCPY(r->dp, a->dp + i, (word32)SP_WORD_SIZEOF * r->used);
+                XMEMCPY(r->dp, a->dp + i,
+                    (word32)sizeof(sp_int_digit) * r->used);
             }
         }
         else {
@@ -15007,7 +15010,7 @@ int sp_mod_2d(const sp_int* a, int e, sp_int* r)
         /* Copy a into r if not same pointer. */
         if (a != r) {
             sp_size_t cnt = (a->used < digits) ? a->used : digits;
-            XMEMCPY(r->dp, a->dp, cnt * (word32)SP_WORD_SIZEOF);
+            XMEMCPY(r->dp, a->dp, cnt * (word32)sizeof(sp_int_digit));
             r->used = a->used;
         #ifdef WOLFSSL_SP_INT_NEGATIVE
             r->sign = a->sign;
@@ -18343,7 +18346,7 @@ int sp_read_unsigned_bin(sp_int* a, const byte* in, word32 inSz)
     }
 
     /* Check a has enough space for number. */
-    if ((err == MP_OKAY) && (inSz > (word32)a->size * SP_WORD_SIZEOF)) {
+    if ((err == MP_OKAY) && (inSz > (word32)a->size * SP_WORD_OCTETS)) {
         err = MP_VAL;
     }
 
@@ -18352,18 +18355,18 @@ int sp_read_unsigned_bin(sp_int* a, const byte* in, word32 inSz)
         int i;
         int j = 0;
 
-        a->used = (sp_size_t)((inSz + SP_WORD_SIZEOF - 1) / SP_WORD_SIZEOF);
+        a->used = (sp_size_t)((inSz + SP_WORD_OCTETS - 1) / SP_WORD_OCTETS);
 
     #if defined(BIG_ENDIAN_ORDER) && !defined(WOLFSSL_SP_INT_DIGIT_ALIGN)
         /* Data endian matches representation of number.
          * Directly copy if we don't have alignment issues.
          */
-        for (i = (int)(inSz-1); i > SP_WORD_SIZEOF-1; i -= SP_WORD_SIZEOF) {
-            a->dp[j++] = *(sp_int_digit*)(in + i - (SP_WORD_SIZEOF - 1));
+        for (i = (int)(inSz-1); i > SP_WORD_OCTETS-1; i -= SP_WORD_OCTETS) {
+            a->dp[j++] = *(sp_int_digit*)(in + i - (SP_WORD_OCTETS - 1));
         }
     #else
         /* Construct digit from required number of bytes. */
-        for (i = (int)(inSz-1); i >= SP_WORD_SIZEOF - 1; i -= SP_WORD_SIZEOF) {
+        for (i = (int)(inSz-1); i >= SP_WORD_OCTETS - 1; i -= SP_WORD_OCTETS) {
             a->dp[j]  = ((sp_int_digit)in[i - 0] <<  0)
         #if SP_WORD_SIZE >= 16
                       | ((sp_int_digit)in[i - 1] <<  8)
@@ -18554,7 +18557,7 @@ int sp_to_unsigned_bin_len_ct(const sp_int* a, byte* out, int outSz)
 
             d = a->dp[i];
             /* Place each byte of a digit into the buffer. */
-            for (b = 0; (j >= 0) && (b < SP_WORD_SIZEOF); b++) {
+            for (b = 0; (j >= 0) && (b < SP_WORD_OCTETS); b++) {
                 out[j--] = (byte)(d & mask);
                 d >>= 8;
             }
@@ -19203,7 +19206,7 @@ int sp_rand_prime(sp_int* r, int len, WC_RNG* rng, void* heap)
         }
 
         /* Get number of digits required to handle required number of bytes. */
-        digits = ((unsigned int)len + SP_WORD_SIZEOF - 1) / SP_WORD_SIZEOF;
+        digits = ((unsigned int)len + SP_WORD_OCTETS - 1) / SP_WORD_OCTETS;
         /* Ensure result has space. */
         if (r->size < digits) {
             err = MP_VAL;
diff --git a/wolfcrypt/test/test.c b/wolfcrypt/test/test.c
index 11414e8850c..c041d38da72 100644
--- a/wolfcrypt/test/test.c
+++ b/wolfcrypt/test/test.c
@@ -93353,7 +93353,7 @@ static wc_test_ret_t mp_test_param(mp_int* a, mp_int* b, mp_int* r, WC_RNG* rng)
     ret = mp_read_unsigned_bin(a, NULL, sizeof(buffer));
     if (ret != WC_NO_ERR_TRACE(MP_VAL))
         return WC_TEST_RET_ENC_EC(ret);
-    ret = mp_read_unsigned_bin(a, buffer, SP_INT_DIGITS * SP_WORD_SIZEOF + 1);
+    ret = mp_read_unsigned_bin(a, buffer, SP_INT_DIGITS * SP_WORD_OCTETS + 1);
     if (ret != WC_NO_ERR_TRACE(MP_VAL))
         return WC_TEST_RET_ENC_EC(ret);
 
diff --git a/wolfssl/wolfcrypt/sp_int.h b/wolfssl/wolfcrypt/sp_int.h
index 4149f1c0e62..6637d70a6b9 100644
--- a/wolfssl/wolfcrypt/sp_int.h
+++ b/wolfssl/wolfcrypt/sp_int.h
@@ -291,8 +291,14 @@ extern "C" {
     #endif
 #endif
 
+/* SP_WORD_OCTETS: number of octets (8-bit units) in an SP digit.
+ * Use for counts of octets in encoded data (e.g. sp_read_unsigned_bin()),
+ * XALIGNED() and preprocessor conditionals - must be an integer literal.
+ * Not the size of a digit in memory where CHAR_BIT != 8 (e.g. TI C28x) - use
+ * sizeof(sp_int_digit) for memory operations (XMEMCPY, XMEMSET, ForceZero).
+ */
 #if SP_WORD_SIZE == 8
-    #define SP_WORD_SIZEOF 1
+    #define SP_WORD_OCTETS 1
 
     typedef   sp_uint8 sp_int_digit;
     typedef    sp_int8 sp_int_sdigit;
@@ -301,7 +307,7 @@ extern "C" {
 
     #define SP_MASK         0xffU
 #elif SP_WORD_SIZE == 16
-    #define SP_WORD_SIZEOF 2
+    #define SP_WORD_OCTETS 2
 
     typedef  sp_uint16 sp_int_digit;
     typedef   sp_int16 sp_int_sdigit;
@@ -310,7 +316,7 @@ extern "C" {
 
     #define SP_MASK         0xffffU
 #elif SP_WORD_SIZE == 32
-    #define SP_WORD_SIZEOF 4
+    #define SP_WORD_OCTETS 4
 
     typedef  sp_uint32 sp_int_digit;
     typedef   sp_int32 sp_int_sdigit;
@@ -319,7 +325,7 @@ extern "C" {
 
     #define SP_MASK         0xffffffffU
 #elif SP_WORD_SIZE == 64
-    #define SP_WORD_SIZEOF 8
+    #define SP_WORD_OCTETS 8
 
     typedef  sp_uint64 sp_int_digit;
     typedef   sp_int64 sp_int_sdigit;
@@ -976,7 +982,7 @@ while (0)
     #define WOLF_BIGINT_DEFINED
 #endif
 
-#if SP_INT_DIGITS < (65536 / SP_WORD_SIZEOF)
+#if SP_INT_DIGITS < (65536 / SP_WORD_OCTETS)
 /* Type for number of digits. */
 typedef word16       sp_size_t;
 #else
@@ -1010,7 +1016,7 @@ typedef struct sp_int {
     struct WC_BIGINT raw;
 #endif
     /** Data of number.  */
-    XALIGNED(SP_WORD_SIZEOF) sp_int_digit dp[SP_INT_DIGITS];
+    XALIGNED(SP_WORD_OCTETS) sp_int_digit dp[SP_INT_DIGITS];
 } sp_int;
 
 typedef struct sp_int_minimal {
@@ -1027,7 +1033,7 @@ typedef struct sp_int_minimal {
     struct WC_BIGINT raw;
 #endif
     /** First digit of number.  */
-    XALIGNED(SP_WORD_SIZEOF) sp_int_digit dp[1];
+    XALIGNED(SP_WORD_OCTETS) sp_int_digit dp[1];
 } sp_int_minimal;
 
 /* MP_INT_SIZEOF_DIGITS() requires that sizeof(sp_int) is a multiple of