diff --git a/crypto/fipsmodule/bn/asm/armv8-mont.pl b/crypto/fipsmodule/bn/asm/armv8-mont.pl
index efee19902..1fce07edd 100644
--- a/crypto/fipsmodule/bn/asm/armv8-mont.pl
+++ b/crypto/fipsmodule/bn/asm/armv8-mont.pl
@@ -55,7 +55,7 @@
  $lo1,$hi1,$nj,$m1,$nlo,$nhi,
  $ovf, $i,$j,$tp,$tj) = map("x$_",6..17,19..24);
 
-# int bn_mul_mont(
+# void bn_mul_mont_nohw(
 $rp="x0";	# BN_ULONG *rp,
 $ap="x1";	# const BN_ULONG *ap,
 $bp="x2";	# const BN_ULONG *bp,
@@ -68,16 +68,11 @@
 
 .text
 
-.globl	bn_mul_mont
-.type	bn_mul_mont,%function
+.globl	bn_mul_mont_nohw
+.type	bn_mul_mont_nohw,%function
 .align	5
-bn_mul_mont:
+bn_mul_mont_nohw:
 	AARCH64_SIGN_LINK_REGISTER
-	tst	$num,#7
-	b.eq	__bn_sqr8x_mont
-	tst	$num,#3
-	b.eq	__bn_mul4x_mont
-.Lmul_mont:
 	stp	x29,x30,[sp,#-64]!
 	add	x29,sp,#0
 	stp	x19,x20,[sp,#16]
@@ -272,7 +267,7 @@
 	ldr	x29,[sp],#64
 	AARCH64_VALIDATE_LINK_REGISTER
 	ret
-.size	bn_mul_mont,.-bn_mul_mont
+.size	bn_mul_mont_nohw,.-bn_mul_mont_nohw
 ___
 {
 ########################################################################
@@ -285,14 +280,11 @@
 my ($tp,$ap_end,$na0)=($bp,$np,$carry);
 
 $code.=<<___;
-.type	__bn_sqr8x_mont,%function
+.globl	bn_sqr8x_mont
+.type	bn_sqr8x_mont,%function
 .align	5
-__bn_sqr8x_mont:
-	// Not adding AARCH64_SIGN_LINK_REGISTER here because __bn_sqr8x_mont is jumped to
-	// only from bn_mul_mont which has already signed the return address.
-	cmp	$ap,$bp
-	b.ne	__bn_mul4x_mont
-.Lsqr8x_mont:
+bn_sqr8x_mont:
+	AARCH64_SIGN_LINK_REGISTER
 	stp	x29,x30,[sp,#-128]!
 	add	x29,sp,#0
 	stp	x19,x20,[sp,#16]
@@ -1049,7 +1041,7 @@
 	// x30 is popped earlier
 	AARCH64_VALIDATE_LINK_REGISTER
 	ret
-.size	__bn_sqr8x_mont,.-__bn_sqr8x_mont
+.size	bn_sqr8x_mont,.-bn_sqr8x_mont
 ___
 }
 
@@ -1068,12 +1060,11 @@
 my  ($carry,$topmost) = ($rp,"x30");
 
 $code.=<<___;
-.type	__bn_mul4x_mont,%function
+.globl	bn_mul4x_mont
+.type	bn_mul4x_mont,%function
 .align	5
-__bn_mul4x_mont:
-	// Not adding AARCH64_SIGN_LINK_REGISTER here because __bn_mul4x_mont is jumped to
-	// only from bn_mul_mont or __bn_mul8x_mont which have already signed the
-	// return address.
+bn_mul4x_mont:
+	AARCH64_SIGN_LINK_REGISTER
 	stp	x29,x30,[sp,#-128]!
 	add	x29,sp,#0
 	stp	x19,x20,[sp,#16]
@@ -1510,7 +1501,7 @@
 	// x30 is popped earlier
 	AARCH64_VALIDATE_LINK_REGISTER
 	ret
-.size	__bn_mul4x_mont,.-__bn_mul4x_mont
+.size	bn_mul4x_mont,.-bn_mul4x_mont
 ___
 }
 $code.=<<___;
diff --git a/crypto/fipsmodule/bn/internal.h b/crypto/fipsmodule/bn/internal.h
index f370257ad..b888fce08 100644
--- a/crypto/fipsmodule/bn/internal.h
+++ b/crypto/fipsmodule/bn/internal.h
@@ -208,6 +208,16 @@ static inline void bn_mul_mont_small(
     const BN_ULONG *np, const BN_ULONG *n0, size_t num) {
     bn_mul_mont_nohw(rp, ap, bp, np, n0, num);
 }
+#elif defined(OPENSSL_AARCH64)
+void bn_mul_mont_nohw(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,
+                      const BN_ULONG *np, const BN_ULONG *n0, size_t num);
+static inline void bn_mul_mont_small(
+    BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,
+    const BN_ULONG *np, const BN_ULONG *n0, size_t num) {
+    // No point in optimizing for P-256 because P-256 doesn't call into
+    // this on AArch64.
+    bn_mul_mont_nohw(rp, ap, bp, np, n0, num);
+}
 #elif defined(OPENSSL_ARM)
 void bn_mul8x_mont_neon(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,
                         const BN_ULONG *np, const BN_ULONG *n0, size_t num);
diff --git a/src/arithmetic.rs b/src/arithmetic.rs
index ec892c9a9..f810741f0 100644
--- a/src/arithmetic.rs
+++ b/src/arithmetic.rs
@@ -17,6 +17,7 @@ use crate::{error::LenMismatchError, limb::LIMB_BITS};
 
 #[macro_use]
 mod ffi;
+mod aarch64_mont;
 mod x86_64_mont;
 
 mod constant;
diff --git a/src/arithmetic/aarch64_mont.rs b/src/arithmetic/aarch64_mont.rs
new file mode 100644
index 000000000..4e88b303e
--- /dev/null
+++ b/src/arithmetic/aarch64_mont.rs
@@ -0,0 +1,59 @@
+// Copyright 2025 Brian Smith.
+//
+// Permission to use, copy, modify, and/or distribute this software for any
+// purpose with or without fee is hereby granted, provided that the above
+// copyright notice and this permission notice appear in all copies.
+//
+// THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
+// WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+// MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY
+// SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+// WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION
+// OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN
+// CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+
+#![cfg(all(target_arch = "aarch64", target_endian = "little"))]
+
+use super::{inout::AliasingSlices3 as _, n0::N0, LimbSliceError, MAX_LIMBS};
+use crate::{
+    c,
+    limb::Limb,
+    polyfill::slice::{AsChunks, AsChunksMut},
+};
+use core::num::NonZeroUsize;
+
+#[inline]
+pub(super) fn sqr_mont5(
+    mut in_out: AsChunksMut<Limb, 8>,
+    n: AsChunks<Limb, 8>,
+    n0: &N0,
+) -> Result<(), LimbSliceError> {
+    prefixed_extern! {
+        // `r` and/or 'a' may alias.
+        // XXX: BoringSSL (kinda, implicitly) declares this to return `int`.
+        // `num` must be a non-zero multiple of 8.
+        fn bn_sqr8x_mont(
+            rp: *mut Limb,
+            ap: *const Limb,
+            ap_again: *const Limb,
+            np: *const Limb,
+            n0: &N0,
+            num: c::NonZero_size_t);
+    }
+
+    let in_out = in_out.as_flattened_mut();
+    let n = n.as_flattened();
+    let num_limbs = NonZeroUsize::new(n.len()).ok_or_else(|| LimbSliceError::too_short(n.len()))?;
+
+    // Avoid stack overflow from the alloca inside.
+    if num_limbs.get() > MAX_LIMBS {
+        return Err(LimbSliceError::too_long(num_limbs.get()));
+    }
+
+    in_out
+        .with_non_dangling_non_null_pointers_rab(num_limbs, |r, a, a_again| {
+            let n = n.as_ptr(); // Non-dangling because num_limbs > 0.
+            unsafe { bn_sqr8x_mont(r, a, a_again, n, n0, num_limbs) };
+        })
+        .map_err(LimbSliceError::len_mismatch)
+}
diff --git a/src/arithmetic/montgomery.rs b/src/arithmetic/montgomery.rs
index 7060e528b..895ef180c 100644
--- a/src/arithmetic/montgomery.rs
+++ b/src/arithmetic/montgomery.rs
@@ -125,9 +125,17 @@ pub(super) fn limbs_mul_mont(
     cfg_if! {
         if #[cfg(all(target_arch = "aarch64", target_endian = "little"))] {
             let _: cpu::Features = cpu;
-            bn_mul_mont_ffi!(in_out, n, n0, (), unsafe {
-                (MIN_LIMBS, MOD_FALLBACK, ()) => bn_mul_mont
-            })
+            const MIN_4X: usize = 4;
+            const MOD_4X: usize = 4;
+            if n.len() >= MIN_4X && n.len() % MOD_4X == 0 {
+                bn_mul_mont_ffi!(in_out, n, n0, (), unsafe {
+                    (MIN_4X, MOD_4X, ()) => bn_mul4x_mont
+                })
+            } else {
+                bn_mul_mont_ffi!(in_out, n, n0, (), unsafe {
+                    (MIN_LIMBS, MOD_FALLBACK, ()) => bn_mul_mont_nohw
+                })
+            }
         } else if #[cfg(all(target_arch = "arm", target_endian = "little"))] {
             const MIN_8X: usize = 8;
             const MOD_8X: usize = 8;
@@ -297,6 +305,15 @@ pub(super) fn limbs_square_mont(
     n0: &N0,
     cpu: cpu::Features,
 ) -> Result<(), LimbSliceError> {
+    #[cfg(all(target_arch = "aarch64", target_endian = "little"))]
+    {
+        use super::aarch64_mont;
+        use crate::polyfill::slice;
+        if let ((r, []), (n, [])) = (slice::as_chunks_mut(r), slice::as_chunks(n)) {
+            return aarch64_mont::sqr_mont5(r, n, n0);
+        }
+    }
+
     #[cfg(target_arch = "x86_64")]
     {
         use super::x86_64_mont;