diff --git a/CHANGELOG.md b/CHANGELOG.md
index f1e3d17f23..57806e4911 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -6,6 +6,10 @@ and this project adheres to
 
 ## [Unreleased]
 
+### Added
+
+- cosmwasm-std: Implement `Sub` and `SubAssign` for `Uint128`
+
 ### Removed
 
 - cosmwasm-std: Make `Uint128` inner field private ([#905])
diff --git a/packages/std/src/math/uint128.rs b/packages/std/src/math/uint128.rs
index 84cd1b7262..ab7ba7d1e5 100644
--- a/packages/std/src/math/uint128.rs
+++ b/packages/std/src/math/uint128.rs
@@ -206,6 +206,22 @@ impl<'a> ops::Add<&'a Uint128> for Uint128 {
     }
 }
 
+impl ops::Sub<Uint128> for Uint128 {
+    type Output = Self;
+
+    fn sub(self, rhs: Self) -> Self {
+        Uint128(self.u128().checked_sub(rhs.u128()).unwrap())
+    }
+}
+
+impl<'a> ops::Sub<&'a Uint128> for Uint128 {
+    type Output = Self;
+
+    fn sub(self, rhs: &'a Uint128) -> Self {
+        Uint128(self.u128().checked_sub(rhs.u128()).unwrap())
+    }
+}
+
 impl ops::AddAssign<Uint128> for Uint128 {
     fn add_assign(&mut self, rhs: Uint128) {
         self.0 = self.0.checked_add(rhs.u128()).unwrap();
@@ -218,6 +234,18 @@ impl<'a> ops::AddAssign<&'a Uint128> for Uint128 {
     }
 }
 
+impl ops::SubAssign<Uint128> for Uint128 {
+    fn sub_assign(&mut self, rhs: Uint128) {
+        self.0 = self.0.checked_sub(rhs.u128()).unwrap();
+    }
+}
+
+impl<'a> ops::SubAssign<&'a Uint128> for Uint128 {
+    fn sub_assign(&mut self, rhs: &'a Uint128) {
+        self.0 = self.0.checked_sub(rhs.u128()).unwrap();
+    }
+}
+
 impl Uint128 {
     /// Returns `self * numerator / denominator`
     pub fn multiply_ratio<A: Into<u128>, B: Into<u128>>(
@@ -397,7 +425,8 @@ mod tests {
         assert_eq!(a + &b, Uint128(35801));
 
         // test - with owned and reference right hand side
-        assert_eq!((b.checked_sub(a)).unwrap(), Uint128(11111));
+        assert_eq!(b - a, Uint128(11111));
+        assert_eq!(b - &a, Uint128(11111));
 
         // test += with owned and reference right hand side
         let mut c = Uint128(300000);
@@ -407,6 +436,14 @@ mod tests {
         d += &b;
         assert_eq!(d, Uint128(323456));
 
+        // test -= with owned and reference right hand side
+        let mut c = Uint128(300000);
+        c -= b;
+        assert_eq!(c, Uint128(276544));
+        let mut d = Uint128(300000);
+        d -= &b;
+        assert_eq!(d, Uint128(276544));
+
         // error result on underflow (- would produce negative result)
         let underflow_result = a.checked_sub(b);
         let OverflowError {
@@ -417,12 +454,18 @@ mod tests {
 
     #[test]
     #[should_panic]
-    fn uint128_math_overflow_panics() {
+    fn uint128_add_overflow_panics() {
         // almost_max is 2^128 - 10
         let almost_max = Uint128(340282366920938463463374607431768211446);
         let _ = almost_max + Uint128(12);
     }
 
+    #[test]
+    #[should_panic]
+    fn uint128_sub_overflow_panics() {
+        let _ = Uint128(1) - Uint128(2);
+    }
+
     #[test]
     fn uint128_multiply_ratio_works() {
         let base = Uint128(500);