diff --git a/src/tests/mod.rs b/src/tests/mod.rs
index e07c3a8..3ad8ba8 100644
--- a/src/tests/mod.rs
+++ b/src/tests/mod.rs
@@ -17,6 +17,19 @@ fn assert_error_properties() {
 	assert_std_err_impl::<Error>();
 }
 
+/// Test that converting an u32 (u64) that does not fit in an i32 (i64)
+/// to a RuntimeValue and back works as expected and the number remains unchanged.
+#[test]
+fn unsigned_to_runtime_value() {
+	use super::RuntimeValue;
+
+	let overflow_i32: u32 = ::std::i32::MAX as u32 + 1;
+	assert_eq!(RuntimeValue::from(overflow_i32).try_into::<u32>().unwrap(), overflow_i32);
+
+	let overflow_i64: u64 = ::std::i64::MAX as u64 + 1;
+	assert_eq!(RuntimeValue::from(overflow_i64).try_into::<u64>().unwrap(), overflow_i64);
+}
+
 pub fn parse_wat(source: &str) -> Module {
 	let wasm_binary = wabt::wat2wasm(source).expect("Failed to parse wat source");
 	Module::from_buffer(wasm_binary).expect("Failed to load parsed module")
diff --git a/src/value.rs b/src/value.rs
index 08cdff2..908829b 100644
--- a/src/value.rs
+++ b/src/value.rs
@@ -1,6 +1,7 @@
 use byteorder::{LittleEndian, ReadBytesExt, WriteBytesExt};
 use nan_preserving_float::{F32, F64};
 use std::io;
+use std::mem::transmute;
 use std::{f32, i32, i64, u32, u64};
 use TrapKind;
 
@@ -188,13 +189,13 @@ impl From<i64> for RuntimeValue {
 
 impl From<u32> for RuntimeValue {
 	fn from(val: u32) -> Self {
-		RuntimeValue::I32(val as i32)
+		RuntimeValue::I32(val.transmute_into())
 	}
 }
 
 impl From<u64> for RuntimeValue {
 	fn from(val: u64) -> Self {
-		RuntimeValue::I64(val as i64)
+		RuntimeValue::I64(val.transmute_into())
 	}
 }
 
@@ -403,11 +404,8 @@ macro_rules! impl_transmute_into_as {
 }
 
 impl_transmute_into_as!(i8, u8);
-impl_transmute_into_as!(u8, i8);
 impl_transmute_into_as!(i32, u32);
-impl_transmute_into_as!(u32, i32);
 impl_transmute_into_as!(i64, u64);
-impl_transmute_into_as!(u64, i64);
 
 macro_rules! impl_transmute_into_npf {
 	($npf:ident, $float:ident, $signed:ident, $unsigned:ident) => {
@@ -468,6 +466,14 @@ impl TransmuteInto<f64> for i64 {
 	fn transmute_into(self) -> f64 { f64::from_bits(self as u64) }
 }
 
+impl TransmuteInto<i32> for u32 {
+	fn transmute_into(self) -> i32 { unsafe { transmute(self) } }
+}
+
+impl TransmuteInto<i64> for u64 {
+	fn transmute_into(self) -> i64 { unsafe { transmute(self) } }
+}
+
 impl LittleEndianConvert for i8 {
 	fn into_little_endian(self) -> Vec<u8> {
 		vec![self as u8]