diff --git a/cranelift/filetests/filetests/runtests/br_icmp_overflow.clif b/cranelift/filetests/filetests/runtests/br_icmp_overflow.clif
new file mode 100644
index 0000000000..71c3a43169
--- /dev/null
+++ b/cranelift/filetests/filetests/runtests/br_icmp_overflow.clif
@@ -0,0 +1,217 @@
+test interpret
+test run
+target aarch64
+target x86_64 machinst
+
+; TODO: Merge this with the main br_icmp file when s390x supports overflows.
+; See: https://github.com/bytecodealliance/wasmtime/issues/3060
+
+function %bricmp_of_i64(i64, i64) -> b1 {
+block0(v0: i64, v1: i64):
+    br_icmp.i64 of v0, v1, block2
+    jump block1
+
+block1:
+    v2 = bconst.b1 false
+    return v2
+
+block2:
+    v3 = bconst.b1 true
+    return v3
+}
+; run: %bricmp_of_i64(0, 0) == false
+; run: %bricmp_of_i64(0, 1) == false
+; run: %bricmp_of_i64(1, 0) == false
+; run: %bricmp_of_i64(0, -1) == false
+; run: %bricmp_of_i64(0x80000000_00000000, 0x80000000_00000000) == false
+; run: %bricmp_of_i64(0x7FFFFFFF_FFFFFFFF, 1) == false
+; run: %bricmp_of_i64(0x7FFFFFFF_FFFFFFFF, 0x7FFFFFFF_FFFFFFFF) == false
+; run: %bricmp_of_i64(0xFFFFFFFF_FFFFFFFF, 1) == false
+; run: %bricmp_of_i64(0x80000000_00000000, 1) == true
+; run: %bricmp_of_i64(0x7FFFFFFF_FFFFFFFF, 0x80000000_00000000) == true
+; run: %bricmp_of_i64(0x80000000_00000000, 0x7FFFFFFF_FFFFFFFF) == true
+; run: %bricmp_of_i64(0x7FFFFFFF_FFFFFFFF, 0xFFFFFFFF_FFFFFFFF) == true
+
+function %bricmp_of_i32(i32, i32) -> b1 {
+block0(v0: i32, v1: i32):
+    br_icmp.i32 of v0, v1, block2
+    jump block1
+
+block1:
+    v2 = bconst.b1 false
+    return v2
+
+block2:
+    v3 = bconst.b1 true
+    return v3
+}
+; run: %bricmp_of_i32(0, 0) == false
+; run: %bricmp_of_i32(0, 1) == false
+; run: %bricmp_of_i32(1, 0) == false
+; run: %bricmp_of_i32(0, -1) == false
+; run: %bricmp_of_i32(0x80000000, 0x80000000) == false
+; run: %bricmp_of_i32(0x7FFFFFFF, 1) == false
+; run: %bricmp_of_i32(0x7FFFFFFF, 0x7FFFFFFF) == false
+; run: %bricmp_of_i32(0xFFFFFFFF, 1) == false
+; run: %bricmp_of_i32(0x80000000, 1) == true
+; run: %bricmp_of_i32(0x7FFFFFFF, 0x80000000) == true
+; run: %bricmp_of_i32(0x80000000, 0x7FFFFFFF) == true
+; run: %bricmp_of_i32(0x7FFFFFFF, 0xFFFFFFFF) == true
+
+function %bricmp_of_i16(i16, i16) -> b1 {
+block0(v0: i16, v1: i16):
+    br_icmp.i16 of v0, v1, block2
+    jump block1
+
+block1:
+    v2 = bconst.b1 false
+    return v2
+
+block2:
+    v3 = bconst.b1 true
+    return v3
+}
+; run: %bricmp_of_i16(0, 0) == false
+; run: %bricmp_of_i16(0, 1) == false
+; run: %bricmp_of_i16(1, 0) == false
+; run: %bricmp_of_i16(0, -1) == false
+; run: %bricmp_of_i16(0x8000, 0x8000) == false
+; run: %bricmp_of_i16(0x7FFF, 1) == false
+; run: %bricmp_of_i16(0x7FFF, 0x7FFF) == false
+; run: %bricmp_of_i16(0xFFFF, 1) == false
+; run: %bricmp_of_i16(0x8000, 1) == true
+; run: %bricmp_of_i16(0x7FFF, 0x8000) == true
+; run: %bricmp_of_i16(0x8000, 0x7FFF) == true
+; run: %bricmp_of_i16(0x7FFF, 0xFFFF) == true
+
+function %bricmp_of_i8(i8, i8) -> b1 {
+block0(v0: i8, v1: i8):
+    br_icmp.i8 of v0, v1, block2
+    jump block1
+
+block1:
+    v2 = bconst.b1 false
+    return v2
+
+block2:
+    v3 = bconst.b1 true
+    return v3
+}
+; run: %bricmp_of_i8(0, 0) == false
+; run: %bricmp_of_i8(0, 1) == false
+; run: %bricmp_of_i8(1, 0) == false
+; run: %bricmp_of_i8(0, -1) == false
+; run: %bricmp_of_i8(0x80, 0x80) == false
+; run: %bricmp_of_i8(0x7F, 1) == false
+; run: %bricmp_of_i8(0x7F, 0x7F) == false
+; run: %bricmp_of_i8(0xFF, 1) == false
+; run: %bricmp_of_i8(0x80, 1) == true
+; run: %bricmp_of_i8(0x7F, 0x80) == true
+; run: %bricmp_of_i8(0x80, 0x7F) == true
+; run: %bricmp_of_i8(0x7F, 0xFF) == true
+
+
+
+function %bricmp_nof_i64(i64, i64) -> b1 {
+block0(v0: i64, v1: i64):
+    br_icmp.i64 nof v0, v1, block2
+    jump block1
+
+block1:
+    v2 = bconst.b1 false
+    return v2
+
+block2:
+    v3 = bconst.b1 true
+    return v3
+}
+; run: %bricmp_nof_i64(0, 0) == true
+; run: %bricmp_nof_i64(0, 1) == true
+; run: %bricmp_nof_i64(1, 0) == true
+; run: %bricmp_nof_i64(0, -1) == true
+; run: %bricmp_nof_i64(0x80000000_00000000, 0x80000000_00000000) == true
+; run: %bricmp_nof_i64(0x7FFFFFFF_FFFFFFFF, 1) == true
+; run: %bricmp_nof_i64(0x7FFFFFFF_FFFFFFFF, 0x7FFFFFFF_FFFFFFFF) == true
+; run: %bricmp_nof_i64(0xFFFFFFFF_FFFFFFFF, 1) == true
+; run: %bricmp_nof_i64(0x80000000_00000000, 1) == false
+; run: %bricmp_nof_i64(0x7FFFFFFF_FFFFFFFF, 0x80000000_00000000) == false
+; run: %bricmp_nof_i64(0x80000000_00000000, 0x7FFFFFFF_FFFFFFFF) == false
+; run: %bricmp_nof_i64(0x7FFFFFFF_FFFFFFFF, 0xFFFFFFFF_FFFFFFFF) == false
+
+function %bricmp_nof_i32(i32, i32) -> b1 {
+block0(v0: i32, v1: i32):
+    br_icmp.i32 nof v0, v1, block2
+    jump block1
+
+block1:
+    v2 = bconst.b1 false
+    return v2
+
+block2:
+    v3 = bconst.b1 true
+    return v3
+}
+; run: %bricmp_nof_i32(0, 0) == true
+; run: %bricmp_nof_i32(0, 1) == true
+; run: %bricmp_nof_i32(1, 0) == true
+; run: %bricmp_nof_i32(0, -1) == true
+; run: %bricmp_nof_i32(0x80000000, 0x80000000) == true
+; run: %bricmp_nof_i32(0x7FFFFFFF, 1) == true
+; run: %bricmp_nof_i32(0x7FFFFFFF, 0x7FFFFFFF) == true
+; run: %bricmp_nof_i32(0xFFFFFFFF, 1) == true
+; run: %bricmp_nof_i32(0x80000000, 1) == false
+; run: %bricmp_nof_i32(0x7FFFFFFF, 0x80000000) == false
+; run: %bricmp_nof_i32(0x80000000, 0x7FFFFFFF) == false
+; run: %bricmp_nof_i32(0x7FFFFFFF, 0xFFFFFFFF) == false
+
+function %bricmp_nof_i16(i16, i16) -> b1 {
+block0(v0: i16, v1: i16):
+    br_icmp.i16 nof v0, v1, block2
+    jump block1
+
+block1:
+    v2 = bconst.b1 false
+    return v2
+
+block2:
+    v3 = bconst.b1 true
+    return v3
+}
+; run: %bricmp_nof_i16(0, 0) == true
+; run: %bricmp_nof_i16(0, 1) == true
+; run: %bricmp_nof_i16(1, 0) == true
+; run: %bricmp_nof_i16(0, -1) == true
+; run: %bricmp_nof_i16(0x8000, 0x8000) == true
+; run: %bricmp_nof_i16(0x7FFF, 1) == true
+; run: %bricmp_nof_i16(0x7FFF, 0x7FFF) == true
+; run: %bricmp_nof_i16(0xFFFF, 1) == true
+; run: %bricmp_nof_i16(0x8000, 1) == false
+; run: %bricmp_nof_i16(0x7FFF, 0x8000) == false
+; run: %bricmp_nof_i16(0x8000, 0x7FFF) == false
+; run: %bricmp_nof_i16(0x7FFF, 0xFFFF) == false
+
+function %bricmp_nof_i8(i8, i8) -> b1 {
+block0(v0: i8, v1: i8):
+    br_icmp.i8 nof v0, v1, block2
+    jump block1
+
+block1:
+    v2 = bconst.b1 false
+    return v2
+
+block2:
+    v3 = bconst.b1 true
+    return v3
+}
+; run: %bricmp_nof_i8(0, 0) == true
+; run: %bricmp_nof_i8(0, 1) == true
+; run: %bricmp_nof_i8(1, 0) == true
+; run: %bricmp_nof_i8(0, -1) == true
+; run: %bricmp_nof_i8(0x80, 0x80) == true
+; run: %bricmp_nof_i8(0x7F, 1) == true
+; run: %bricmp_nof_i8(0x7F, 0x7F) == true
+; run: %bricmp_nof_i8(0xFF, 1) == true
+; run: %bricmp_nof_i8(0x80, 1) == false
+; run: %bricmp_nof_i8(0x7F, 0x80) == false
+; run: %bricmp_nof_i8(0x80, 0x7F) == false
+; run: %bricmp_nof_i8(0x7F, 0xFF) == false
diff --git a/cranelift/interpreter/src/step.rs b/cranelift/interpreter/src/step.rs
index 28699b8273..2843be21f6 100644
--- a/cranelift/interpreter/src/step.rs
+++ b/cranelift/interpreter/src/step.rs
@@ -720,8 +720,8 @@ where
             &left.clone().convert(ValueConversionKind::ToUnsigned)?,
             &right.clone().convert(ValueConversionKind::ToUnsigned)?,
         )?,
-        IntCC::Overflow => unimplemented!("IntCC::Overflow"),
-        IntCC::NotOverflow => unimplemented!("IntCC::NotOverflow"),
+        IntCC::Overflow => Value::of(left, right)?,
+        IntCC::NotOverflow => !Value::of(left, right)?,
     })
 }
 
diff --git a/cranelift/interpreter/src/value.rs b/cranelift/interpreter/src/value.rs
index bf3e5e2691..492a7e5fef 100644
--- a/cranelift/interpreter/src/value.rs
+++ b/cranelift/interpreter/src/value.rs
@@ -37,6 +37,7 @@ pub trait Value: Clone + From<DataValue> {
         Ok(other.eq(self)? || other.gt(self)?)
     }
     fn uno(&self, other: &Self) -> ValueResult<bool>;
+    fn of(&self, other: &Self) -> ValueResult<bool>;
 
     // Arithmetic.
     fn add(self, other: Self) -> ValueResult<Self>;
@@ -277,6 +278,16 @@ impl Value for DataValue {
         Ok(self.is_nan()? || other.is_nan()?)
     }
 
+    fn of(&self, other: &Self) -> ValueResult<bool> {
+        Ok(match (self, other) {
+            (DataValue::I8(a), DataValue::I8(b)) => a.checked_sub(*b).is_none(),
+            (DataValue::I16(a), DataValue::I16(b)) => a.checked_sub(*b).is_none(),
+            (DataValue::I32(a), DataValue::I32(b)) => a.checked_sub(*b).is_none(),
+            (DataValue::I64(a), DataValue::I64(b)) => a.checked_sub(*b).is_none(),
+            _ => unimplemented!(),
+        })
+    }
+
     fn add(self, other: Self) -> ValueResult<Self> {
         binary_match!(wrapping_add(&self, &other); [I8, I16, I32, I64]) // TODO: floats must handle NaNs, +/-0
     }