diff --git a/layers/endian.go b/layers/endian.go
new file mode 100644
index 000000000..76c4a6b90
--- /dev/null
+++ b/layers/endian.go
@@ -0,0 +1,23 @@
+package layers
+
+import (
+	"encoding/binary"
+	"unsafe"
+)
+
+var endian binary.ByteOrder = _endian()
+
+// _endian returns the binary.ByteOrder as defined for the host byte order
+func _endian() binary.ByteOrder {
+	buf := [2]byte{}
+	*(*uint16)(unsafe.Pointer(&buf[0])) = uint16(0xABCD)
+
+	switch buf {
+	case [2]byte{0xCD, 0xAB}:
+		return binary.LittleEndian
+	case [2]byte{0xAB, 0xCD}:
+		return binary.BigEndian
+	default:
+		panic("Could not determine native endianness.")
+	}
+}
diff --git a/layers/ip4.go b/layers/ip4.go
index 2b3c0c6bf..fda23581f 100644
--- a/layers/ip4.go
+++ b/layers/ip4.go
@@ -112,7 +112,13 @@ func (ip *IPv4) SerializeTo(b gopacket.SerializeBuffer, opts gopacket.SerializeO
 	}
 	bytes[0] = (ip.Version << 4) | ip.IHL
 	bytes[1] = ip.TOS
-	binary.BigEndian.PutUint16(bytes[2:], ip.Length)
+
+	if opts.IPLengthHostByteOrder {
+		endian.PutUint16(bytes[2:], ip.Length)
+	} else {
+		binary.BigEndian.PutUint16(bytes[2:], ip.Length)
+	}
+
 	binary.BigEndian.PutUint16(bytes[4:], ip.Id)
 	binary.BigEndian.PutUint16(bytes[6:], ip.flagsfrags())
 	bytes[8] = ip.TTL
diff --git a/layers/ip6.go b/layers/ip6.go
index 87b9d33d5..302dad3bb 100644
--- a/layers/ip6.go
+++ b/layers/ip6.go
@@ -206,7 +206,13 @@ func (ipv6 *IPv6) SerializeTo(b gopacket.SerializeBuffer, opts gopacket.Serializ
 			ipv6.Length = uint16(pLen)
 		}
 	}
-	binary.BigEndian.PutUint16(bytes[4:], ipv6.Length)
+
+	if opts.IPLengthHostByteOrder {
+		endian.PutUint16(bytes[4:], ipv6.Length)
+	} else {
+		binary.BigEndian.PutUint16(bytes[4:], ipv6.Length)
+	}
+
 	bytes[6] = byte(ipv6.NextHeader)
 	bytes[7] = byte(ipv6.HopLimit)
 	if err := ipv6.AddressTo16(); err != nil {
diff --git a/writer.go b/writer.go
index 5d303dc4a..a57199ec3 100644
--- a/writer.go
+++ b/writer.go
@@ -46,6 +46,10 @@ type SerializeOptions struct {
 	// FixLengths determines whether, during serialization, layers should fix
 	// the values for any length field that depends on the payload.
 	FixLengths bool
+	// IPLengthHostByteOrder determines whether, during serialization, the network
+	// layer containing the IPv4/v6 header length field should be serialized in
+	// host byte order.
+	IPLengthHostByteOrder bool
 	// ComputeChecksums determines whether, during serialization, layers
 	// should recompute checksums based on their payloads.
 	ComputeChecksums bool
@@ -69,9 +73,9 @@ type SerializeOptions struct {
 // byte slices returned by any previous Bytes() call (the same buffer is
 // reused).
 //
-//  1) Reusing a write buffer is generally much faster than creating a new one,
+//  1. Reusing a write buffer is generally much faster than creating a new one,
 //     and with the default implementation it avoids additional memory allocations.
-//  2) If a byte slice from a previous Bytes() call will continue to be used,
+//  2. If a byte slice from a previous Bytes() call will continue to be used,
 //     it's better to create a new SerializeBuffer.
 //
 // The Clear method is specifically designed to minimize memory allocations for
@@ -197,12 +201,13 @@ func (w *serializeBuffer) PushLayer(l LayerType) {
 // invalidates all slices previously returned by w.Bytes()
 //
 // Example:
-//   buf := gopacket.NewSerializeBuffer()
-//   opts := gopacket.SerializeOptions{}
-//   gopacket.SerializeLayers(buf, opts, a, b, c)
-//   firstPayload := buf.Bytes()  // contains byte representation of a(b(c))
-//   gopacket.SerializeLayers(buf, opts, d, e, f)
-//   secondPayload := buf.Bytes()  // contains byte representation of d(e(f)). firstPayload is now invalidated, since the SerializeLayers call Clears buf.
+//
+//	buf := gopacket.NewSerializeBuffer()
+//	opts := gopacket.SerializeOptions{}
+//	gopacket.SerializeLayers(buf, opts, a, b, c)
+//	firstPayload := buf.Bytes()  // contains byte representation of a(b(c))
+//	gopacket.SerializeLayers(buf, opts, d, e, f)
+//	secondPayload := buf.Bytes()  // contains byte representation of d(e(f)). firstPayload is now invalidated, since the SerializeLayers call Clears buf.
 func SerializeLayers(w SerializeBuffer, opts SerializeOptions, layers ...SerializableLayer) error {
 	w.Clear()
 	for i := len(layers) - 1; i >= 0; i-- {