diff --git a/Cargo.toml b/Cargo.toml
index 5b22f8b..c0f7e2d 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -20,13 +20,15 @@ ark-relations = { git = "https://github.com/arkworks-rs/snark", default-features
 
 derivative = { version = "2", features = ["use_core"] }
 tracing = { version = "0.1", default-features = false, features = [ "attributes" ] }
+num-bigint = {version = "0.3", default-features = false }
+num-traits = {version = "0.2", default-features = false }
 
 [dev-dependencies]
 rand = { version = "0.7", default-features = false }
 rand_xorshift = "0.2"
-ark-test-curves = { git = "https://github.com/arkworks-rs/algebra", default-features = false, features = ["bls12_381_scalar_field"] }
+ark-test-curves = { git = "https://github.com/arkworks-rs/algebra", default-features = false, features = ["bls12_381_scalar_field", "mnt4_753_scalar_field"] }
 
 [features]
 default = ["std"]
-std = [ "ark-ff/std", "ark-relations/std", "ark-std/std" ]
+std = [ "ark-ff/std", "ark-relations/std", "ark-std/std", "num-bigint/std" ]
 parallel = [ "std", "ark-ff/parallel" ]
diff --git a/src/bits/mod.rs b/src/bits/mod.rs
index ccedc02..803c0d0 100644
--- a/src/bits/mod.rs
+++ b/src/bits/mod.rs
@@ -17,6 +17,7 @@ pub mod uint;
 make_uint!(UInt16, 16, u16, uint16, "16");
 make_uint!(UInt32, 32, u32, uint32, "32");
 make_uint!(UInt64, 64, u64, uint64, "64");
+make_uint!(UInt128, 128, u128, uint128, "128");
 
 /// Specifies constraints for conversion to a little-endian bit representation
 /// of `self`.
diff --git a/src/bits/uint.rs b/src/bits/uint.rs
index 40c9d95..40ff267 100644
--- a/src/bits/uint.rs
+++ b/src/bits/uint.rs
@@ -6,9 +6,11 @@ macro_rules! make_uint {
         #[doc = $native_doc_name]
         #[doc = "`type."]
         pub mod $mod_name {
-            use ark_ff::{Field, FpParameters, PrimeField};
+            use ark_ff::{Field, FpParameters, One, PrimeField, Zero};
             use core::borrow::Borrow;
             use core::convert::TryFrom;
+            use num_bigint::BigUint;
+            use num_traits::cast::ToPrimitive;
 
             use ark_relations::r1cs::{
                 ConstraintSystemRef, LinearCombination, Namespace, SynthesisError, Variable,
@@ -183,6 +185,9 @@ macro_rules! make_uint {
                     // in the scalar field
                     assert!(F::Params::MODULUS_BITS >= 2 * $size);
 
+                    // Support up to 128
+                    assert!($size <= 128);
+
                     assert!(operands.len() >= 1);
                     assert!($size * operands.len() <= F::Params::MODULUS_BITS as usize);
 
@@ -192,10 +197,10 @@ macro_rules! make_uint {
 
                     // Compute the maximum value of the sum so we allocate enough bits for
                     // the result
-                    let mut max_value = (operands.len() as u128) * u128::from($native::max_value());
+                    let mut max_value = BigUint::from($native::max_value()) * BigUint::from(operands.len());
 
                     // Keep track of the resulting value
-                    let mut result_value = Some(0u128);
+                    let mut result_value = Some(BigUint::zero());
 
                     // This is a linear combination that we will enforce to be "zero"
                     let mut lc = LinearCombination::zero();
@@ -207,7 +212,7 @@ macro_rules! make_uint {
                         // Accumulate the value
                         match op.value {
                             Some(val) => {
-                                result_value.as_mut().map(|v| *v += u128::from(val));
+                                result_value.as_mut().map(|v| *v += BigUint::from(val));
                             }
 
                             None => {
@@ -246,7 +251,12 @@ macro_rules! make_uint {
                     }
 
                     // The value of the actual result is modulo 2^$size
-                    let modular_value = result_value.map(|v| v as $native);
+                    let modular_value = result_value.clone().map(|v|
+                        {
+                            let modulus = BigUint::from(1u64) << ($size as u32);
+                            (v % modulus).to_u128().unwrap() as $native
+                        }
+                    );
 
                     if all_constants && modular_value.is_some() {
                         // We can just return a constant, rather than
@@ -262,11 +272,9 @@ macro_rules! make_uint {
                     // Allocate each bit_gadget of the result
                     let mut coeff = F::one();
                     let mut i = 0;
-                    while max_value != 0 {
+                    while max_value != BigUint::zero() {
                         // Allocate the bit_gadget
-                        let b = AllocatedBit::new_witness(cs.clone(), || {
-                            result_value.map(|v| (v >> i) & 1 == 1).get()
-                        })?;
+                        let b = AllocatedBit::new_witness(cs.clone(), || result_value.clone().map(|v| (v >> i) & BigUint::one() == BigUint::one()).get())?;
 
                         // Subtract this bit_gadget from the linear combination to ensure the sums
                         // balance out
@@ -384,7 +392,7 @@ macro_rules! make_uint {
             mod test {
                 use super::$name;
                 use crate::{bits::boolean::Boolean, prelude::*, Vec};
-                use ark_test_curves::bls12_381::Fr;
+                use ark_test_curves::mnt4_753::Fr;
                 use ark_relations::r1cs::{ConstraintSystem, SynthesisError};
                 use rand::{Rng, SeedableRng};
                 use rand_xorshift::XorShiftRng;
@@ -517,7 +525,6 @@ macro_rules! make_uint {
                         let r = $name::addmany(&[r, c_bit, d_bit]).unwrap();
 
                         assert!(cs.is_satisfied().unwrap());
-
                         assert!(r.value == Some(expected));
 
                         for b in r.bits.iter() {