diff --git a/src/constants.rs b/src/constants.rs
index 1821f9a..50879da 100644
--- a/src/constants.rs
+++ b/src/constants.rs
@@ -166,6 +166,13 @@ pub const l_minus_1: Scalar = Scalar([ 0xec, 0xd3, 0xf5, 0x5c, 0x1a, 0x63, 0x12,
                                        0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
                                        0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10 ]);
 
+/// `lminus1` is the order of base point minus two, i.e. 2^252 +
+/// 27742317777372353535851937790883648493 - 2, in little-endian form
+pub const l_minus_2: Scalar = Scalar([ 0xeb, 0xd3, 0xf5, 0x5c, 0x1a, 0x63, 0x12, 0x58,
+                                       0xd6, 0x9c, 0xf7, 0xa2, 0xde, 0xf9, 0xde, 0x14,
+                                       0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+                                       0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10 ]);
+
 /// The 8-torsion subgroup Ɛ[8].
 ///
 /// In the case of Curve25519, it is cyclic; the `i`th element of the
diff --git a/src/scalar.rs b/src/scalar.rs
index 6468ddd..d711173 100644
--- a/src/scalar.rs
+++ b/src/scalar.rs
@@ -209,6 +209,11 @@ impl Scalar {
                  0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ])
     }
 
+    /// Compute the multiplicative inverse of this scalar.
+    pub fn invert(&self) -> Scalar {
+        self.unpack().invert().pack()
+    }
+
     /// Get the bits of the scalar.
     pub fn bits(&self) -> [i8;256] {
         let mut bits = [0i8; 256];
@@ -462,6 +467,19 @@ impl UnpackedScalar {
         UnpackedScalar([1,0,0,0,0,0,0,0,0,0,0,0])
     }
 
+    /// Compute the multiplicative inverse of this scalar.
+    pub fn invert(&self) -> UnpackedScalar {
+        let mut y = UnpackedScalar::one();
+        // Run through bits of l-2 from highest to least
+        for bit in constants::l_minus_2.bits().iter().rev() {
+            y = &y * &y;
+            if *bit == 1 {
+                y *= self;
+            }
+        }
+        y
+    }
+
     /// Compute `ab+c (mod l)`.
     pub fn multiply_add(a: &UnpackedScalar,
                         b: &UnpackedScalar,
@@ -727,6 +745,14 @@ mod test {
         }
     }
 
+    #[test]
+    fn invert() {
+        let x = UnpackedScalar([2,0,0,0,0,0,0,0,0,0,0,0]);
+        let x_inv = x.invert();
+        let should_be_one = &x * &x_inv;
+        assert_eq!(should_be_one.pack(), Scalar::one());
+    }
+
     // Negating a scalar twice should result in the original scalar.
     #[test]
     fn scalar_neg() {
@@ -757,6 +783,12 @@ mod bench {
         b.iter(|| Scalar::multiply_add(&X, &Y, &Z) );
     }
 
+    #[bench]
+    fn invert(b: &mut Bencher) {
+        let x = X.unpack();
+        b.iter(|| x.invert());
+    }
+
     #[bench]
     fn scalar_unpacked_multiply_add(b: &mut Bencher) {
         let x = X.unpack();