diff --git a/CHANGELOG.md b/CHANGELOG.md
index 6ab63036c2..94f19a0ee1 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -11,8 +11,13 @@ and this project adheres to
 - cosmwasm-std: Implement `&T + T` and `&T op &T` for `Uint64`, `Uint128`,
   `Uint256` and `Uint512`; improve panic message for `Uint64::add` and
   `Uint512::add` ([#2092])
+- cosmwasm-std: Add `Uint{64,128,256,512}::panicking_add` and `::panicking_sub`
+  which are like the `Add`/`Sub` implementations but `const`. ([#2098])
+- cosmwasm-std: Let `Timestamp::plus_nanos`/`::minus_nanos` use
+  `Uint64::panicking_add`/`::panicking_sub` and document overflows. ([#2098])
 
 [#2092]: https://github.com/CosmWasm/cosmwasm/pull/2092
+[#2098]: https://github.com/CosmWasm/cosmwasm/pull/2098
 
 ### Changed
 
diff --git a/packages/std/src/math/uint128.rs b/packages/std/src/math/uint128.rs
index db59d469d3..70bdf11c54 100644
--- a/packages/std/src/math/uint128.rs
+++ b/packages/std/src/math/uint128.rs
@@ -259,6 +259,28 @@ impl Uint128 {
         Self(self.0.saturating_pow(exp))
     }
 
+    /// This is the same as [`Uint128::add`] but const.
+    ///
+    /// Panics on overflow.
+    #[must_use = "this returns the result of the operation, without modifying the original"]
+    pub const fn panicking_add(self, other: Self) -> Self {
+        match self.0.checked_add(other.u128()) {
+            None => panic!("attempt to add with overflow"),
+            Some(sum) => Self(sum),
+        }
+    }
+
+    /// This is the same as [`Uint128::sub`] but const.
+    ///
+    /// Panics on overflow.
+    #[must_use = "this returns the result of the operation, without modifying the original"]
+    pub const fn panicking_sub(self, other: Self) -> Self {
+        match self.0.checked_sub(other.u128()) {
+            None => panic!("attempt to subtract with overflow"),
+            Some(diff) => Self(diff),
+        }
+    }
+
     #[must_use = "this returns the result of the operation, without modifying the original"]
     pub const fn abs_diff(self, other: Self) -> Self {
         Self(if self.0 < other.0 {
@@ -363,11 +385,7 @@ impl Add<Uint128> for Uint128 {
     type Output = Self;
 
     fn add(self, rhs: Self) -> Self {
-        Self(
-            self.u128()
-                .checked_add(rhs.u128())
-                .expect("attempt to add with overflow"),
-        )
+        self.panicking_add(rhs)
     }
 }
 forward_ref_binop!(impl Add, add for Uint128, Uint128);
@@ -376,11 +394,7 @@ impl Sub<Uint128> for Uint128 {
     type Output = Self;
 
     fn sub(self, rhs: Self) -> Self {
-        Uint128(
-            self.u128()
-                .checked_sub(rhs.u128())
-                .expect("attempt to subtract with overflow"),
-        )
+        self.panicking_sub(rhs)
     }
 }
 forward_ref_binop!(impl Sub, sub for Uint128, Uint128);
@@ -1148,6 +1162,21 @@ mod tests {
         assert_eq!(a, Uint128::from(1u32));
     }
 
+    #[test]
+    fn uint128_panicking_sub_works() {
+        let a = Uint128::new(5);
+        let b = Uint128::new(3);
+        assert_eq!(a.panicking_sub(b), Uint128::new(2));
+    }
+
+    #[test]
+    #[should_panic(expected = "attempt to subtract with overflow")]
+    fn uint128_panicking_sub_panics_on_overflow() {
+        let a = Uint128::zero();
+        let b = Uint128::one();
+        let _diff = a.panicking_sub(b);
+    }
+
     #[test]
     fn uint128_abs_diff_works() {
         let a = Uint128::from(42u32);
diff --git a/packages/std/src/math/uint256.rs b/packages/std/src/math/uint256.rs
index 3e1d55971e..22749837b6 100644
--- a/packages/std/src/math/uint256.rs
+++ b/packages/std/src/math/uint256.rs
@@ -329,6 +329,28 @@ impl Uint256 {
         Self(self.0.saturating_pow(exp))
     }
 
+    /// This is the same as [`Uint256::add`] but const.
+    ///
+    /// Panics on overflow.
+    #[must_use = "this returns the result of the operation, without modifying the original"]
+    pub const fn panicking_add(self, other: Self) -> Self {
+        match self.0.checked_add(other.0) {
+            None => panic!("attempt to add with overflow"),
+            Some(sum) => Self(sum),
+        }
+    }
+
+    /// This is the same as [`Uint256::sub`] but const.
+    ///
+    /// Panics on overflow.
+    #[must_use = "this returns the result of the operation, without modifying the original"]
+    pub const fn panicking_sub(self, other: Self) -> Self {
+        match self.0.checked_sub(other.0) {
+            None => panic!("attempt to subtract with overflow"),
+            Some(diff) => Self(diff),
+        }
+    }
+
     #[must_use = "this returns the result of the operation, without modifying the original"]
     pub const fn abs_diff(self, other: Self) -> Self {
         Self(self.0.abs_diff(other.0))
@@ -428,11 +450,7 @@ impl Add<Uint256> for Uint256 {
     type Output = Self;
 
     fn add(self, rhs: Self) -> Self {
-        Self(
-            self.0
-                .checked_add(rhs.0)
-                .expect("attempt to add with overflow"),
-        )
+        self.panicking_add(rhs)
     }
 }
 forward_ref_binop!(impl Add, add for Uint256, Uint256);
@@ -441,11 +459,7 @@ impl Sub<Uint256> for Uint256 {
     type Output = Self;
 
     fn sub(self, rhs: Self) -> Self {
-        Self(
-            self.0
-                .checked_sub(rhs.0)
-                .expect("attempt to subtract with overflow"),
-        )
+        self.panicking_sub(rhs)
     }
 }
 forward_ref_binop!(impl Sub, sub for Uint256, Uint256);
@@ -1690,6 +1704,21 @@ mod tests {
         assert_eq!(a, Uint256::from(1u32));
     }
 
+    #[test]
+    fn uint256_panicking_sub_works() {
+        let a = Uint256::from(5u32);
+        let b = Uint256::from(3u32);
+        assert_eq!(a.panicking_sub(b), Uint256::from(2u32));
+    }
+
+    #[test]
+    #[should_panic(expected = "attempt to subtract with overflow")]
+    fn uint256_panicking_sub_panics_on_overflow() {
+        let a = Uint256::zero();
+        let b = Uint256::one();
+        let _diff = a.panicking_sub(b);
+    }
+
     #[test]
     fn uint256_abs_diff_works() {
         let a = Uint256::from(42u32);
diff --git a/packages/std/src/math/uint512.rs b/packages/std/src/math/uint512.rs
index a6de3592b8..53b1fc7892 100644
--- a/packages/std/src/math/uint512.rs
+++ b/packages/std/src/math/uint512.rs
@@ -294,6 +294,28 @@ impl Uint512 {
         Self(self.0.saturating_pow(exp))
     }
 
+    /// This is the same as [`Uint512::add`] but const.
+    ///
+    /// Panics on overflow.
+    #[must_use = "this returns the result of the operation, without modifying the original"]
+    pub const fn panicking_add(self, other: Self) -> Self {
+        match self.0.checked_add(other.0) {
+            None => panic!("attempt to add with overflow"),
+            Some(sum) => Self(sum),
+        }
+    }
+
+    /// This is the same as [`Uint512::sub`] but const.
+    ///
+    /// Panics on overflow.
+    #[must_use = "this returns the result of the operation, without modifying the original"]
+    pub const fn panicking_sub(self, other: Self) -> Self {
+        match self.0.checked_sub(other.0) {
+            None => panic!("attempt to subtract with overflow"),
+            Some(diff) => Self(diff),
+        }
+    }
+
     #[must_use = "this returns the result of the operation, without modifying the original"]
     pub const fn abs_diff(self, other: Self) -> Self {
         Self(self.0.abs_diff(other.0))
@@ -415,11 +437,7 @@ impl Add<Uint512> for Uint512 {
     type Output = Self;
 
     fn add(self, rhs: Self) -> Self {
-        Self(
-            self.0
-                .checked_add(rhs.0)
-                .expect("attempt to add with overflow"),
-        )
+        self.panicking_add(rhs)
     }
 }
 forward_ref_binop!(impl Add, add for Uint512, Uint512);
@@ -428,7 +446,7 @@ impl Sub<Uint512> for Uint512 {
     type Output = Self;
 
     fn sub(self, rhs: Self) -> Self {
-        Uint512(self.0.checked_sub(rhs.0).unwrap())
+        self.panicking_sub(rhs)
     }
 }
 forward_ref_binop!(impl Sub, sub for Uint512, Uint512);
@@ -1338,6 +1356,21 @@ mod tests {
         assert_eq!(a, Uint512::from(1u32));
     }
 
+    #[test]
+    fn uint512_panicking_sub_works() {
+        let a = Uint512::from(5u32);
+        let b = Uint512::from(3u32);
+        assert_eq!(a.panicking_sub(b), Uint512::from(2u32));
+    }
+
+    #[test]
+    #[should_panic(expected = "attempt to subtract with overflow")]
+    fn uint512_panicking_sub_panics_on_overflow() {
+        let a = Uint512::zero();
+        let b = Uint512::one();
+        let _diff = a.panicking_sub(b);
+    }
+
     #[test]
     fn uint512_abs_diff_works() {
         let a = Uint512::from(42u32);
diff --git a/packages/std/src/math/uint64.rs b/packages/std/src/math/uint64.rs
index e71abd977a..a865c8a95c 100644
--- a/packages/std/src/math/uint64.rs
+++ b/packages/std/src/math/uint64.rs
@@ -252,6 +252,28 @@ impl Uint64 {
         Self(self.0.saturating_pow(exp))
     }
 
+    /// This is the same as [`Uint64::add`] but const.
+    ///
+    /// Panics on overflow.
+    #[must_use = "this returns the result of the operation, without modifying the original"]
+    pub const fn panicking_add(self, other: Self) -> Self {
+        match self.0.checked_add(other.u64()) {
+            None => panic!("attempt to add with overflow"),
+            Some(sum) => Self(sum),
+        }
+    }
+
+    /// This is the same as [`Uint64::sub`] but const.
+    ///
+    /// Panics on overflow.
+    #[must_use = "this returns the result of the operation, without modifying the original"]
+    pub const fn panicking_sub(self, other: Self) -> Self {
+        match self.0.checked_sub(other.u64()) {
+            None => panic!("attempt to subtract with overflow"),
+            Some(diff) => Self(diff),
+        }
+    }
+
     #[must_use = "this returns the result of the operation, without modifying the original"]
     pub const fn abs_diff(self, other: Self) -> Self {
         Self(if self.0 < other.0 {
@@ -326,11 +348,7 @@ impl Add<Uint64> for Uint64 {
     type Output = Self;
 
     fn add(self, rhs: Self) -> Self {
-        Self(
-            self.u64()
-                .checked_add(rhs.u64())
-                .expect("attempt to add with overflow"),
-        )
+        self.panicking_add(rhs)
     }
 }
 forward_ref_binop!(impl Add, add for Uint64, Uint64);
@@ -339,11 +357,7 @@ impl Sub<Uint64> for Uint64 {
     type Output = Self;
 
     fn sub(self, rhs: Self) -> Self {
-        Uint64(
-            self.u64()
-                .checked_sub(rhs.u64())
-                .expect("attempt to subtract with overflow"),
-        )
+        self.panicking_sub(rhs)
     }
 }
 forward_ref_binop!(impl Sub, sub for Uint64, Uint64);
@@ -1062,6 +1076,21 @@ mod tests {
         assert_eq!(a, Uint64::from(1u32));
     }
 
+    #[test]
+    fn uint64_panicking_sub_works() {
+        let a = Uint64::new(5);
+        let b = Uint64::new(3);
+        assert_eq!(a.panicking_sub(b), Uint64::new(2));
+    }
+
+    #[test]
+    #[should_panic(expected = "attempt to subtract with overflow")]
+    fn uint64_panicking_sub_panics_on_overflow() {
+        let a = Uint64::zero();
+        let b = Uint64::one();
+        let _diff = a.panicking_sub(b);
+    }
+
     #[test]
     fn uint64_abs_diff_works() {
         let a = Uint64::from(42u32);
diff --git a/packages/std/src/timestamp.rs b/packages/std/src/timestamp.rs
index 6a10f060d5..70167b8285 100644
--- a/packages/std/src/timestamp.rs
+++ b/packages/std/src/timestamp.rs
@@ -38,62 +38,105 @@ impl Timestamp {
         Timestamp(Uint64::new(seconds_since_epoch * 1_000_000_000))
     }
 
+    /// Adds the given amount of days to the timestamp and
+    /// returns the result. The original value remains unchanged.
+    ///
+    /// Panics if the result exceeds the value range of [`Timestamp`].
     #[must_use = "this returns the result of the operation, without modifying the original"]
     #[inline]
     pub const fn plus_days(&self, addition: u64) -> Timestamp {
         self.plus_hours(addition * 24)
     }
 
+    /// Adds the given amount of hours to the timestamp and
+    /// returns the result. The original value remains unchanged.
+    ///
+    /// Panics if the result exceeds the value range of [`Timestamp`].
     #[must_use = "this returns the result of the operation, without modifying the original"]
     #[inline]
     pub const fn plus_hours(&self, addition: u64) -> Timestamp {
         self.plus_minutes(addition * 60)
     }
 
+    /// Adds the given amount of minutes to the timestamp and
+    /// returns the result. The original value remains unchanged.
+    ///
+    /// Panics if the result exceeds the value range of [`Timestamp`].
     #[must_use = "this returns the result of the operation, without modifying the original"]
     #[inline]
     pub const fn plus_minutes(&self, addition: u64) -> Timestamp {
         self.plus_seconds(addition * 60)
     }
 
+    /// Adds the given amount of seconds to the timestamp and
+    /// returns the result. The original value remains unchanged.
+    ///
+    /// Panics if the result exceeds the value range of [`Timestamp`].
     #[must_use = "this returns the result of the operation, without modifying the original"]
+    #[inline]
     pub const fn plus_seconds(&self, addition: u64) -> Timestamp {
         self.plus_nanos(addition * 1_000_000_000)
     }
 
+    /// Adds the given amount of nanoseconds to the timestamp and
+    /// returns the result. The original value remains unchanged.
+    ///
+    /// Panics if the result exceeds the value range of [`Timestamp`].
     #[must_use = "this returns the result of the operation, without modifying the original"]
+    // no #[inline] here as this could be shared with all the callers
     pub const fn plus_nanos(&self, addition: u64) -> Timestamp {
-        let nanos = Uint64::new(self.0.u64() + addition);
+        let nanos = self.0.panicking_add(Uint64::new(addition));
         Timestamp(nanos)
     }
 
+    /// Subtracts the given amount of days from the timestamp and
+    /// returns the result. The original value remains unchanged.
+    ///
+    /// Panics if the result is not >= 0. I.e. times before epoch cannot be represented.
     #[must_use = "this returns the result of the operation, without modifying the original"]
     #[inline]
     pub const fn minus_days(&self, subtrahend: u64) -> Timestamp {
         self.minus_hours(subtrahend * 24)
     }
 
+    /// Subtracts the given amount of hours from the timestamp and
+    /// returns the result. The original value remains unchanged.
+    ///
+    /// Panics if the result is not >= 0. I.e. times before epoch cannot be represented.
     #[must_use = "this returns the result of the operation, without modifying the original"]
     #[inline]
     pub const fn minus_hours(&self, subtrahend: u64) -> Timestamp {
         self.minus_minutes(subtrahend * 60)
     }
 
+    /// Subtracts the given amount of minutes from the timestamp and
+    /// returns the result. The original value remains unchanged.
+    ///
+    /// Panics if the result is not >= 0. I.e. times before epoch cannot be represented.
     #[must_use = "this returns the result of the operation, without modifying the original"]
     #[inline]
     pub const fn minus_minutes(&self, subtrahend: u64) -> Timestamp {
         self.minus_seconds(subtrahend * 60)
     }
 
+    /// Subtracts the given amount of seconds from the timestamp and
+    /// returns the result. The original value remains unchanged.
+    ///
+    /// Panics if the result is not >= 0. I.e. times before epoch cannot be represented.
     #[must_use = "this returns the result of the operation, without modifying the original"]
+    #[inline]
     pub const fn minus_seconds(&self, subtrahend: u64) -> Timestamp {
         self.minus_nanos(subtrahend * 1_000_000_000)
     }
 
+    /// Subtracts the given amount of nanoseconds from the timestamp and
+    /// returns the result. The original value remains unchanged.
+    ///
+    /// Panics if the result is not >= 0. I.e. times before epoch cannot be represented.
     #[must_use = "this returns the result of the operation, without modifying the original"]
+    // no #[inline] here as this could be shared with all the callers
     pub const fn minus_nanos(&self, subtrahend: u64) -> Timestamp {
-        let nanos = Uint64::new(self.0.u64() - subtrahend);
-        Timestamp(nanos)
+        Timestamp(self.0.panicking_sub(Uint64::new(subtrahend)))
     }
 
     /// Returns nanoseconds since epoch
@@ -160,6 +203,13 @@ mod tests {
         assert_eq!(sum.0.u64(), 123);
     }
 
+    #[test]
+    #[should_panic(expected = "attempt to add with overflow")]
+    fn timestamp_plus_nanos_panics_on_overflow() {
+        let max = Timestamp::from_nanos(u64::MAX);
+        let _earlier = max.plus_nanos(20);
+    }
+
     #[test]
     fn timestamp_minus_seconds() {
         let earlier = Timestamp::from_seconds(123).minus_seconds(0);