diff --git a/src/codex32/bip93.py b/src/codex32/bip93.py
index 02afad7..0b77fc7 100644
--- a/src/codex32/bip93.py
+++ b/src/codex32/bip93.py
@@ -93,6 +93,8 @@ def _checksum_for_encoded_length(hrp: str, encoded_length: int) -> _Checksum:
 
 def _decode_codex32(text: str) -> tuple[str, _ProfileRules | None, tuple[int, ...], _Checksum]:
     hrp, encoded = bech32_decode(text)
+    if len(hrp) > 83:
+        raise InvalidLength(f"human-readable part exceeds 83 characters ({len(hrp)})")
     if len(hrp) + 1 + len(encoded) < 21:
         raise InvalidLength("codex32 string must contain at least 21 characters")
     checksum = _checksum_for_encoded_length(hrp, len(encoded))
diff --git a/src/codex32/correction.py b/src/codex32/correction.py
index 4a1a561..111d5b0 100644
--- a/src/codex32/correction.py
+++ b/src/codex32/correction.py
@@ -763,6 +763,8 @@ def _validate_context(context: CorrectionContext) -> None:
         _validate_single_case_ascii(context.hrp)
         if context.hrp.lower() != context.hrp:
             raise ValueError("hrp must be a normalized application prefix")
+        if len(context.hrp) > 83:
+            raise ValueError(f"human-readable part exceeds 83 characters ({len(context.hrp)})")
         length = context.expected_length
         if length is not None:
             if isinstance(length, bool) or not isinstance(length, int):
diff --git a/tests/test_correction_bch.py b/tests/test_correction_bch.py
index 49b8800..9e47b6c 100644
--- a/tests/test_correction_bch.py
+++ b/tests/test_correction_bch.py
@@ -217,6 +217,7 @@ def test_public_context_constrains_length_prefix_and_used_indices() -> None:
         CorrectionContext(Profile.MS, excluded_indices=["a"]),  # type: ignore[arg-type]
         CorrectionContext(Profile.MS, excluded_indices=("s",)),
         CorrectionContext(Profile.MS, excluded_indices=("a", "A")),
+        CorrectionContext("z" * 84, expected_length=106),
     ),
 )
 def test_malformed_public_context_is_rejected(context: CorrectionContext) -> None:
diff --git a/tests/test_generic_hrp.py b/tests/test_generic_hrp.py
index 341e504..9245cb1 100644
--- a/tests/test_generic_hrp.py
+++ b/tests/test_generic_hrp.py
@@ -18,7 +18,7 @@
     recover_secret,
 )
 from codex32.cli import main, ms_main
-from codex32.errors import MismatchedHrp, MismatchedProfile
+from codex32.errors import InvalidLength, MismatchedHrp, MismatchedProfile
 
 UNKNOWN = {
     "short": {
@@ -50,6 +50,14 @@ def test_opaque_hrp_parse_complete_recover_and_derive(vector: dict[str, str]) ->
     assert derive_share([secret, a], "d").text == vector["D"].upper()
 
 
+def test_hrp_over_83_characters_is_rejected_at_the_bip173_limit() -> None:
+    within_limit = oracle_encode("z" * 83, "0testsq")
+    over_limit = oracle_encode("z" * 84, "0testsq")
+    assert parse_codex32(within_limit).hrp == "z" * 83
+    with pytest.raises(InvalidLength):
+        parse_codex32(over_limit)
+
+
 def test_sharing_compares_normalized_hrp_and_keeps_compatibility_error_name() -> None:
     assert MismatchedProfile is MismatchedHrp
     zz = parse_codex32(UNKNOWN["short"]["S"])