diff --git a/include/instructions.h b/include/instructions.h
index 2b29aa5..32ec884 100644
--- a/include/instructions.h
+++ b/include/instructions.h
@@ -36,6 +36,7 @@ public:
   void and_rm32_r32();  // 0x21
   void and_r32_rm32();  // 0x23
   void and_eax_imm32(); // 0x25
+  void sub_rm32_r32();  // 0x29
   void xor_rm32_r32();  // 0x31
   void inc_eax();       // 0x40
   void inc_ecx();       // 0x41
diff --git a/instructions.cpp b/instructions.cpp
index 79caf39..a4ef651 100644
--- a/instructions.cpp
+++ b/instructions.cpp
@@ -29,6 +29,7 @@ void Instructions::init_instructions(){
   this->instructions[0x21] = &Instructions::and_rm32_r32;
   this->instructions[0x23] = &Instructions::and_r32_rm32;
   this->instructions[0x25] = &Instructions::and_eax_imm32;
+  this->instructions[0x29] = &Instructions::sub_rm32_r32;
   this->instructions[0x31] = &Instructions::xor_rm32_r32;
   this->instructions[0x40] = &Instructions::inc_eax;
   this->instructions[0x41] = &Instructions::inc_ecx;
@@ -423,6 +424,55 @@ void Instructions::and_eax_imm32(){
   this->registers[0] &= imm32;
 }
 
+void Instructions::sub_rm32_r32(){
+  //printf("sub_rm32_r32 called.\n");
+  uint32_t addr, dst, imm32;
+  uint8_t imm8;
+
+  this->modrm = memory.read_uint8(this->eip);
+  this->calc_modrm();
+
+  switch (this->mod) {
+    case 0:
+      // sub [M], R
+      // addr : M
+      this->eip++;
+      addr = this->registers[this->M];
+      // dst : data of [M]
+      dst = memory.read_uint32(addr);
+      memory.write_uint32(addr, dst - this->registers[this->R]);
+      break;
+    case 1:
+      // sub [M+imm8], R
+      this->eip++;
+      imm8 = memory.read_uint8(this->eip);
+      // addr : M
+      addr = this->registers[this->M];
+      // dst : data of [M+imm8]
+      dst = memory.read_uint32(addr + imm8);
+      memory.write_uint32(addr + imm8, dst - this->registers[this->R]);
+      break;
+    case 2:
+      // sub [M+imm32], R
+      this->eip++;
+      imm32 = memory.read_uint32(this->eip);
+      imm32 = swap_endian32(imm32);
+      // addr : M
+      addr = this->registers[this->M];
+      // dst : data of [M+imm32]
+      dst = memory.read_uint32(addr + imm32);
+      memory.write_uint32(addr, dst - this->registers[this->R]);
+      this->eip += 4;
+      break;
+    default:
+      // case mod == 3
+      // sub M, R
+      this->eip++;
+      this->registers[this->M] -= this->registers[this->R];
+      break;
+  }
+}
+
 void Instructions::xor_rm32_r32(){
   //printf("xor_rm32_r32 called.\n");
   uint32_t addr, dst, imm32;