diff --git a/cpu.cpp b/cpu.cpp
index e5a3cfb..1f01104 100644
--- a/cpu.cpp
+++ b/cpu.cpp
@@ -41,6 +41,7 @@ void cpu::emulation_start()
 	instruction_count = 0;
 
 	running_since = get_ms();
+	wait_time     = 0;
 }
 
 bool cpu::check_breakpoint()
@@ -75,7 +76,7 @@ std::tuple<double, double, uint64_t> cpu::get_mips_rel_speed()
 {
 	uint64_t instr_count = get_instructions_executed_count();
 
-        uint32_t t_diff = get_ms() - running_since;  // TODO fix this because we now implement WAIT where it sits idle
+        uint32_t t_diff = get_ms() - running_since - (wait_time / 1000);
 
         double mips = instr_count / (1000.0 * t_diff);
 
@@ -1563,6 +1564,7 @@ bool cpu::misc_operations(const uint16_t instr)
 
 		case 0b0000000000000001: // WAIT
 			{
+				uint64_t start = get_us();
 #if defined(BUILD_FOR_RP2040)
 				uint8_t rc = 0;
 				xQueueReceive(qi_q, &rc, 0);
@@ -1571,6 +1573,9 @@ bool cpu::misc_operations(const uint16_t instr)
 
 				qi_cv.wait(lck);
 #endif
+				uint64_t end = get_us();
+
+				wait_time += end - start;  // used for MIPS calculation
 			}
 
 			DOLOG(debug, false, "WAIT returned");
diff --git a/cpu.h b/cpu.h
index 9683dda..e06f125 100644
--- a/cpu.h
+++ b/cpu.h
@@ -31,15 +31,16 @@ class cpu
 private:
 	uint16_t regs0_5[2][6]; // R0...5, selected by bit 11 in PSW, 
 	uint16_t sp[3 + 1]; // stackpointers, MF../MT.. select via 12/13 from PSW, others via 14/15
-	uint16_t pc    { 0 };
-	uint16_t psw   { 0 };
-	uint16_t fpsr  { 0 };
-	uint16_t stackLimitRegister { 0377 };
-	int      processing_trap_depth { 0 };
-	uint64_t instruction_count { 0     };
-	uint64_t running_since     { 0     };
-	bool     it_is_a_trap      { false };
-	uint64_t mtpi_count        { 0     };
+	uint16_t pc                 { 0     };
+	uint16_t psw                { 0     };
+	uint16_t fpsr               { 0     };
+	uint16_t stackLimitRegister { 0377  };
+	int      processing_trap_depth { 0  };
+	uint64_t instruction_count  { 0     };
+	uint64_t running_since      { 0     };
+	uint64_t wait_time          { 0     };
+	bool     it_is_a_trap       { false };
+	uint64_t mtpi_count         { 0     };
 
 	// level, vector
 	std::map<uint8_t, std::set<uint8_t> > queued_interrupts;