diff --git a/Source/Core/Common/MathUtil.h b/Source/Core/Common/MathUtil.h
index 18a0c305ce..2aa2b406ab 100644
--- a/Source/Core/Common/MathUtil.h
+++ b/Source/Core/Common/MathUtil.h
@@ -6,6 +6,7 @@
 
 #include <algorithm>
 #include <cmath>
+#include <limits>
 #include <type_traits>
 #include <vector>
 
@@ -30,6 +31,48 @@ constexpr auto Lerp(const T& x, const T& y, const F& a) -> decltype(x + (y - x)
   return x + (y - x) * a;
 }
 
+// Casts the specified value to a Dest. The value will be clamped to fit in the destination type.
+// Warning: The result of SaturatingCast(NaN) is undefined.
+template <typename Dest, typename T>
+constexpr Dest SaturatingCast(T value)
+{
+  static_assert(std::is_integral<Dest>());
+
+  constexpr Dest lo = std::numeric_limits<Dest>::lowest();
+  constexpr Dest hi = std::numeric_limits<Dest>::max();
+
+  // T being a signed integer and Dest unsigned is a problematic case because the value will
+  // be converted into an unsigned integer, and u32(...) < 0 is always false.
+  if constexpr (std::is_integral<T>() && std::is_signed<T>() && std::is_unsigned<Dest>())
+  {
+    static_assert(lo == 0);
+    if (value < 0)
+      return lo;
+    // Now that we got rid of negative values, we can safely cast value to an unsigned T
+    // since unsigned T can represent any positive value signed T could represent.
+    // The compiler will then promote the LHS or the RHS if necessary.
+    if (std::make_unsigned_t<T>(value) > hi)
+      return hi;
+  }
+  else if constexpr (std::is_integral<T>() && std::is_unsigned<T>() && std::is_signed<Dest>())
+  {
+    // value and hi will never be negative, and hi is representable as an unsigned Dest.
+    if (value > std::make_unsigned_t<Dest>(hi))
+      return hi;
+  }
+  else
+  {
+    // Do not use std::clamp or a similar function here to avoid overflow.
+    // For example, if Dest = s64 and T = int, we want integer promotion to convert value to a s64
+    // instead of changing lo or hi into an int.
+    if (value < lo)
+      return lo;
+    if (value > hi)
+      return hi;
+  }
+  return static_cast<Dest>(value);
+}
+
 template <typename T>
 constexpr bool IsPow2(T imm)
 {
diff --git a/Source/UnitTests/Common/MathUtilTest.cpp b/Source/UnitTests/Common/MathUtilTest.cpp
index c0609bbb92..e31777d16d 100644
--- a/Source/UnitTests/Common/MathUtilTest.cpp
+++ b/Source/UnitTests/Common/MathUtilTest.cpp
@@ -26,3 +26,45 @@ TEST(MathUtil, NextPowerOf2)
   EXPECT_EQ(8U, MathUtil::NextPowerOf2(6));
   EXPECT_EQ(0x40000000U, MathUtil::NextPowerOf2(0x23456789));
 }
+
+TEST(MathUtil, SaturatingCast)
+{
+  // Cast from an integer type to a smaller type
+  EXPECT_EQ(255u, (MathUtil::SaturatingCast<u8, int>(1000)));
+  EXPECT_EQ(255u, (MathUtil::SaturatingCast<u8, u16>(1000u)));
+  EXPECT_EQ(255u, (MathUtil::SaturatingCast<u8, std::size_t>(1000)));
+
+  // Cast from a signed integer type
+  EXPECT_EQ(0u, (MathUtil::SaturatingCast<u8, int>(-1)));
+  EXPECT_EQ(0u, (MathUtil::SaturatingCast<u8, int>(-1000)));
+  EXPECT_EQ(0u, (MathUtil::SaturatingCast<u32, int>(-1)));
+  EXPECT_EQ(-1000, (MathUtil::SaturatingCast<s16, int>(-1000)));
+  EXPECT_EQ(-1000, (MathUtil::SaturatingCast<int, int>(-1000)));
+  EXPECT_EQ(-1000, (MathUtil::SaturatingCast<s64, int>(-1000)));
+
+  // Cast from an unsigned integer type to a smaller integer type
+  EXPECT_EQ(0x7fff, (MathUtil::SaturatingCast<s16, u32>(0xffffffffu)));
+  EXPECT_EQ(0x7fffffff, (MathUtil::SaturatingCast<int, u32>(0xffffffffu)));
+
+  // Cast from a floating point type to an integer type
+  EXPECT_EQ(255u, MathUtil::SaturatingCast<u8>(1234.0));
+  EXPECT_EQ(0u, MathUtil::SaturatingCast<u8>(-1234.0));
+  EXPECT_EQ(127, MathUtil::SaturatingCast<s8>(5678.0));
+  EXPECT_EQ(-128, MathUtil::SaturatingCast<s8>(-5678.0));
+  EXPECT_EQ(65535u, MathUtil::SaturatingCast<u16>(999999.0));
+
+  // Negative zero
+  EXPECT_EQ(0u, MathUtil::SaturatingCast<u8>(0.0));
+  EXPECT_EQ(0u, MathUtil::SaturatingCast<u8>(-0.0));
+  EXPECT_EQ(0, MathUtil::SaturatingCast<s8>(0.0));
+  EXPECT_EQ(0, MathUtil::SaturatingCast<s8>(-0.0));
+
+  // Edge cases
+  EXPECT_EQ(std::numeric_limits<s32>::max(),
+            MathUtil::SaturatingCast<s32>(std::numeric_limits<float>::infinity()));
+  EXPECT_EQ(std::numeric_limits<s32>::min(),
+            MathUtil::SaturatingCast<s32>(-std::numeric_limits<float>::infinity()));
+  // 16777217 = 2^24 + 1 is the first integer that cannot be represented correctly with a f32.
+  EXPECT_EQ(16777216, MathUtil::SaturatingCast<s32>(float(16777216)));
+  EXPECT_EQ(16777216, MathUtil::SaturatingCast<s32>(float(16777217)));
+}