diff --git a/include/dynarmic/A64/config.h b/include/dynarmic/A64/config.h
index 3bfdb2f0..8cad3e92 100644
--- a/include/dynarmic/A64/config.h
+++ b/include/dynarmic/A64/config.h
@@ -209,18 +209,6 @@ struct UserConfig {
     /// This enables the fast dispatcher.
     bool enable_fast_dispatch = true;
 
-    // The below options relate to accuracy of floating-point emulation.
-
-    /// Determines how accurate NaN handling is.
-    enum class NaNAccuracy {
-        /// Results of operations with NaNs will exactly match hardware.
-        Accurate,
-        /// Behave as if FPCR.DN is always set.
-        AlwaysForceDefaultNaN,
-        /// No special handling of NaN, other than setting default NaN when FPCR.DN is set.
-        NoChecks,
-    } floating_point_nan_accuracy = NaNAccuracy::Accurate;
-
     // Determines whether AddTicks and GetTicksRemaining are called.
     // If false, execution will continue until soon after Jit::HaltExecution is called.
     // bool enable_ticks = true; // TODO
diff --git a/src/backend/x64/a64_emit_x64.cpp b/src/backend/x64/a64_emit_x64.cpp
index 06188e2c..fe9b6c41 100644
--- a/src/backend/x64/a64_emit_x64.cpp
+++ b/src/backend/x64/a64_emit_x64.cpp
@@ -52,10 +52,6 @@ FP::FPCR A64EmitContext::FPCR(bool fpcr_controlled) const {
     return fpcr_controlled ? Location().FPCR() : Location().FPCR().ASIMDStandardValue();
 }
 
-bool A64EmitContext::AccurateNaN() const {
-    return conf.floating_point_nan_accuracy == A64::UserConfig::NaNAccuracy::Accurate;
-}
-
 A64EmitX64::A64EmitX64(BlockOfCode& code, A64::UserConfig conf, A64::Jit* jit_interface)
         : EmitX64(code), conf(conf), jit_interface{jit_interface} {
     GenMemory128Accessors();
diff --git a/src/backend/x64/a64_emit_x64.h b/src/backend/x64/a64_emit_x64.h
index 3817d631..03d57cb3 100644
--- a/src/backend/x64/a64_emit_x64.h
+++ b/src/backend/x64/a64_emit_x64.h
@@ -28,7 +28,6 @@ struct A64EmitContext final : public EmitContext {
     A64::LocationDescriptor Location() const;
     bool IsSingleStep() const;
     FP::FPCR FPCR(bool fpcr_controlled = true) const override;
-    bool AccurateNaN() const override;
 
     const A64::UserConfig& conf;
 };
diff --git a/src/backend/x64/emit_x64.h b/src/backend/x64/emit_x64.h
index 23a0f957..f6cb486b 100644
--- a/src/backend/x64/emit_x64.h
+++ b/src/backend/x64/emit_x64.h
@@ -50,7 +50,6 @@ struct EmitContext {
     void EraseInstruction(IR::Inst* inst);
 
     virtual FP::FPCR FPCR(bool fpcr_controlled = true) const = 0;
-    virtual bool AccurateNaN() const { return true; }
 
     RegAlloc& reg_alloc;
     IR::Block& block;
diff --git a/src/backend/x64/emit_x64_floating_point.cpp b/src/backend/x64/emit_x64_floating_point.cpp
index 5437dc2d..21b13481 100644
--- a/src/backend/x64/emit_x64_floating_point.cpp
+++ b/src/backend/x64/emit_x64_floating_point.cpp
@@ -258,7 +258,7 @@ void FPTwoOp(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst, Function fn) {
 
     Xbyak::Xmm result = ctx.reg_alloc.UseScratchXmm(args[0]);
 
-    if (ctx.AccurateNaN() && !ctx.FPCR().DN()) {
+    if (!ctx.FPCR().DN()) {
         end = ProcessNaN<fsize>(code, result);
     }
     if constexpr (std::is_member_function_pointer_v<Function>) {
@@ -268,7 +268,7 @@ void FPTwoOp(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst, Function fn) {
     }
     if (ctx.FPCR().DN()) {
         ForceToDefaultNaN<fsize>(code, result);
-    } else if (ctx.AccurateNaN()) {
+    } else {
         PostProcessNaN<fsize>(code, result, ctx.reg_alloc.ScratchXmm());
     }
     code.L(end);
@@ -282,7 +282,7 @@ void FPThreeOp(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst, Function fn)
 
     auto args = ctx.reg_alloc.GetArgumentInfo(inst);
 
-    if (ctx.FPCR().DN() || !ctx.AccurateNaN()) {
+    if (ctx.FPCR().DN()) {
         const Xbyak::Xmm result = ctx.reg_alloc.UseScratchXmm(args[0]);
         const Xbyak::Xmm operand = ctx.reg_alloc.UseScratchXmm(args[1]);
 
@@ -292,9 +292,7 @@ void FPThreeOp(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst, Function fn)
             fn(result, operand);
         }
 
-        if (ctx.AccurateNaN()) {
-            ForceToDefaultNaN<fsize>(code, result);
-        }
+        ForceToDefaultNaN<fsize>(code, result);
 
         ctx.reg_alloc.DefineValue(inst, result);
         return;
@@ -437,7 +435,7 @@ static void EmitFPMinMax(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst) {
     code.jmp(end);
 
     code.L(nan);
-    if (ctx.FPCR().DN() || !ctx.AccurateNaN()) {
+    if (ctx.FPCR().DN()) {
         code.movaps(result, code.MConst(xword, fsize == 32 ? f32_nan : f64_nan));
         code.jmp(end);
     } else {
@@ -677,7 +675,7 @@ static void EmitFPMulX(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst) {
 
     auto args = ctx.reg_alloc.GetArgumentInfo(inst);
 
-    const bool do_default_nan = ctx.FPCR().DN() || !ctx.AccurateNaN();
+    const bool do_default_nan = ctx.FPCR().DN();
 
     const Xbyak::Xmm op1 = ctx.reg_alloc.UseXmm(args[0]);
     const Xbyak::Xmm op2 = ctx.reg_alloc.UseXmm(args[1]);
diff --git a/src/backend/x64/emit_x64_vector_floating_point.cpp b/src/backend/x64/emit_x64_vector_floating_point.cpp
index c71de89f..ba3f75e3 100644
--- a/src/backend/x64/emit_x64_vector_floating_point.cpp
+++ b/src/backend/x64/emit_x64_vector_floating_point.cpp
@@ -284,7 +284,7 @@ void EmitTwoOpVectorOperation(BlockOfCode& code, EmitContext& ctx, IR::Inst* ins
     auto args = ctx.reg_alloc.GetArgumentInfo(inst);
     const bool fpcr_controlled = args[fpcr_controlled_arg_index].GetImmediateU1();
 
-    if (!ctx.AccurateNaN() || ctx.FPCR(fpcr_controlled).DN()) {
+    if (ctx.FPCR(fpcr_controlled).DN()) {
         Xbyak::Xmm result;
 
         if constexpr (std::is_member_function_pointer_v<Function>) {
@@ -336,7 +336,7 @@ void EmitThreeOpVectorOperation(BlockOfCode& code, EmitContext& ctx, IR::Inst* i
     auto args = ctx.reg_alloc.GetArgumentInfo(inst);
     const bool fpcr_controlled = args[2].GetImmediateU1();
 
-    if (!ctx.AccurateNaN() || ctx.FPCR(fpcr_controlled).DN()) {
+    if (ctx.FPCR(fpcr_controlled).DN()) {
         const Xbyak::Xmm xmm_a = ctx.reg_alloc.UseScratchXmm(args[0]);
         const Xbyak::Xmm xmm_b = ctx.reg_alloc.UseXmm(args[1]);