diff --git a/mysql-test/main/func_math.result b/mysql-test/main/func_math.result
index 475485cbcae59..179b49e14d36c 100644
--- a/mysql-test/main/func_math.result
+++ b/mysql-test/main/func_math.result
@@ -3739,3 +3739,99 @@ cast(1 as unsigned) - cast(2 as unsigned)
 -1
 set sql_mode=default;
 # End of 10.5 tests
+#
+# MDEV-38670 Minus string is converted to -0
+#
+# Testing arithmetic operations with signed zero
+SELECT -0.0 + 0.0;
+-0.0 + 0.0
+0.0
+SELECT 0.0 + -0.0;
+0.0 + -0.0
+0.0
+SELECT -0.0 + -0.0;
+-0.0 + -0.0
+0.0
+SELECT -0.0 - 0.0;
+-0.0 - 0.0
+0.0
+SELECT  0.0 - '';
+0.0 - ''
+0
+Warnings:
+Warning	1292	Truncated incorrect DOUBLE value: ''
+SELECT -0.0;
+-0.0
+0.0
+SELECT 0.0 * -1.0;
+0.0 * -1.0
+0.00
+SELECT 0e0*-1e0;
+0e0*-1e0
+0
+SELECT -'';
+-''
+0
+Warnings:
+Warning	1292	Truncated incorrect DOUBLE value: ''
+SELECT ABS(-0.0);
+ABS(-0.0)
+0.0
+SELECT -0.0 % 1.0;
+-0.0 % 1.0
+0.0
+# Testing comparison functions with signed zero
+SELECT CASE WHEN -0.0=0.0 THEN 1.0 ELSE 0.0 END;
+CASE WHEN -0.0=0.0 THEN 1.0 ELSE 0.0 END
+1.0
+SELECT CASE WHEN -''=0.0 THEN 1.0 ELSE 0.0 END;
+CASE WHEN -''=0.0 THEN 1.0 ELSE 0.0 END
+1.0
+Warnings:
+Warning	1292	Truncated incorrect DOUBLE value: ''
+SELECT CASE WHEN 0e0*-1e0=0.0 THEN 1.0 ELSE 0.0 END;
+CASE WHEN 0e0*-1e0=0.0 THEN 1.0 ELSE 0.0 END
+1.0
+# Testing with table data
+CREATE TABLE t1 (a DOUBLE, b DOUBLE);
+INSERT INTO t1 VALUES
+(0.0, 1.0),
+(-0.0, 1.0),
+(1.0, 0.0),
+(1.0, -0.0),
+(0.0, 0.0),
+(-0.0, -0.0);
+SELECT a, b, a + b AS sum_ab, a - b AS diff_ab, a * b AS mul_ab FROM t1;
+a	b	sum_ab	diff_ab	mul_ab
+0	1	1	-1	0
+0	1	1	-1	0
+1	0	1	1	0
+1	0	1	1	0
+0	0	0	0	0
+0	0	0	0	0
+SELECT a, b, IFNULL(a, b) AS ifnull_ab, NULLIF(a, b) AS nullif_ab FROM t1;
+a	b	ifnull_ab	nullif_ab
+0	1	0	0
+0	1	0	0
+1	0	1	1
+1	0	1	1
+0	0	0	NULL
+0	0	0	NULL
+SELECT a, b, CASE WHEN a > 0 THEN a ELSE b END AS case_ab FROM t1;
+a	b	case_ab
+0	1	1
+0	1	1
+1	0	1
+1	0	1
+0	0	0
+0	0	0
+SELECT a, b, COALESCE(a, b) AS coalesce_ab FROM t1;
+a	b	coalesce_ab
+0	1	0
+0	1	0
+1	0	1
+1	0	1
+0	0	0
+0	0	0
+DROP TABLE t1;
+# End of 13 tests
diff --git a/mysql-test/main/func_math.test b/mysql-test/main/func_math.test
index 363d84eab3e07..1b26cc6e1a983 100644
--- a/mysql-test/main/func_math.test
+++ b/mysql-test/main/func_math.test
@@ -1998,3 +1998,57 @@ select cast(1 as unsigned) - cast(2 as unsigned);
 set sql_mode=default;
 
 --echo # End of 10.5 tests
+
+--echo #
+--echo # MDEV-38670 Minus string is converted to -0
+--echo #
+
+--source include/have_innodb.inc
+
+# Test arithmetic operations
+--echo # Testing arithmetic operations with signed zero
+
+SELECT -0.0 + 0.0;
+SELECT 0.0 + -0.0;
+SELECT -0.0 + -0.0;
+
+SELECT -0.0 - 0.0;
+SELECT  0.0 - '';
+
+SELECT -0.0;
+SELECT 0.0 * -1.0;
+SELECT 0e0*-1e0;
+SELECT -'';
+
+
+SELECT ABS(-0.0);
+
+SELECT -0.0 % 1.0;
+
+# Test comparison functions
+--echo # Testing comparison functions with signed zero
+
+SELECT CASE WHEN -0.0=0.0 THEN 1.0 ELSE 0.0 END;
+SELECT CASE WHEN -''=0.0 THEN 1.0 ELSE 0.0 END;
+SELECT CASE WHEN 0e0*-1e0=0.0 THEN 1.0 ELSE 0.0 END;
+
+
+# Test with table data
+--echo # Testing with table data
+
+CREATE TABLE t1 (a DOUBLE, b DOUBLE);
+INSERT INTO t1 VALUES
+  (0.0, 1.0),
+  (-0.0, 1.0),
+  (1.0, 0.0),
+  (1.0, -0.0),
+  (0.0, 0.0),
+  (-0.0, -0.0);
+
+SELECT a, b, a + b AS sum_ab, a - b AS diff_ab, a * b AS mul_ab FROM t1;
+SELECT a, b, IFNULL(a, b) AS ifnull_ab, NULLIF(a, b) AS nullif_ab FROM t1;
+SELECT a, b, CASE WHEN a > 0 THEN a ELSE b END AS case_ab FROM t1;
+SELECT a, b, COALESCE(a, b) AS coalesce_ab FROM t1;
+
+DROP TABLE t1;
+--echo # End of 13 tests
diff --git a/sql/item_cmpfunc.cc b/sql/item_cmpfunc.cc
index 24ff774e59cc3..7f8c0810fde85 100644
--- a/sql/item_cmpfunc.cc
+++ b/sql/item_cmpfunc.cc
@@ -2496,7 +2496,7 @@ Item_func_ifnull::real_op()
   value= args[1]->val_real();
   if ((null_value=args[1]->null_value))
     return 0.0;
-  return value;
+  return normalize_signed_zero(value);
 }
 
 longlong
@@ -3036,7 +3036,7 @@ Item_func_nullif::real_op()
   }
   value= args[2]->val_real();
   null_value= args[2]->null_value;
-  return value;
+  return normalize_signed_zero(value);
 }
 
 longlong
@@ -3250,7 +3250,7 @@ double Item_func_case::real_op()
   }
   res= item->val_real();
   null_value=item->null_value;
-  return res;
+  return normalize_signed_zero(res);
 }
 
 
@@ -3649,7 +3649,7 @@ double Item_func_coalesce::real_op()
   {
     double res= args[i]->val_real();
     if (!args[i]->null_value)
-      return res;
+      return normalize_signed_zero(res);
   }
   null_value=1;
   return 0;
diff --git a/sql/item_func.cc b/sql/item_func.cc
index 0d2b43d100844..69f5e4c050546 100644
--- a/sql/item_func.cc
+++ b/sql/item_func.cc
@@ -1172,7 +1172,7 @@ double Item_func_plus::real_op()
   double value= args[0]->val_real() + args[1]->val_real();
   if ((null_value=args[0]->null_value || args[1]->null_value))
     return 0.0;
-  return check_float_overflow(value);
+  return normalize_signed_zero(check_float_overflow(value));
 }
 
 #if defined(__powerpc64__) && GCC_VERSION >= 6003 && GCC_VERSION <= 10002
@@ -1338,7 +1338,7 @@ double Item_func_minus::real_op()
   double value= args[0]->val_real() - args[1]->val_real();
   if ((null_value=args[0]->null_value || args[1]->null_value))
     return 0.0;
-  return check_float_overflow(value);
+  return normalize_signed_zero(check_float_overflow(value));
 }
 
 
@@ -1440,7 +1440,7 @@ double Item_func_mul::real_op()
   double value= args[0]->val_real() * args[1]->val_real();
   if ((null_value=args[0]->null_value || args[1]->null_value))
     return 0.0;
-  return check_float_overflow(value);
+  return normalize_signed_zero(check_float_overflow(value));
 }
 
 
@@ -1522,7 +1522,7 @@ double Item_func_div::real_op()
     signal_divide_by_null();
     return 0.0;
   }
-  return check_float_overflow(value/val2);
+  return normalize_signed_zero(check_float_overflow(value/val2));
 }
 
 
@@ -1712,7 +1712,7 @@ double Item_func_mod::real_op()
     signal_divide_by_null();
     return 0.0;
   }
-  return fmod(value,val2);
+  return normalize_signed_zero(fmod(value,val2));
 }
 
 
@@ -1825,7 +1825,7 @@ double Item_func_neg::real_op()
 {
   double value= args[0]->val_real();
   null_value= args[0]->null_value;
-  return -value;
+  return normalize_signed_zero(-value);
 }
 
 
@@ -1931,7 +1931,7 @@ double Item_func_abs::real_op()
 {
   double value= args[0]->val_real();
   null_value= args[0]->null_value;
-  return fabs(value);
+  return normalize_signed_zero(fabs(value));
 }
 
 
@@ -2409,7 +2409,7 @@ double Item_func_ceiling::real_op()
   */
   volatile double value= args[0]->val_real();
   null_value= args[0]->null_value;
-  return ceil(value);
+  return normalize_signed_zero(ceil(value));
 }
 
 
@@ -2473,7 +2473,7 @@ double Item_func_floor::real_op()
   */
   volatile double value= args[0]->val_real();
   null_value= args[0]->null_value;
-  return floor(value);
+  return normalize_signed_zero(floor(value));
 }
 
 
@@ -2741,7 +2741,7 @@ double Item_func_round::real_op()
   {
     longlong dec= args[1]->val_int();
     if (!(null_value= args[1]->null_value))
-      return my_double_round(value, dec, args[1]->unsigned_flag, truncate);
+      return normalize_signed_zero(my_double_round(value, dec, args[1]->unsigned_flag, truncate));
   }
   return 0.0;
 }
diff --git a/sql/item_func.h b/sql/item_func.h
index 0be8cf61591b1..a6c89a6b5163f 100644
--- a/sql/item_func.h
+++ b/sql/item_func.h
@@ -285,6 +285,10 @@ class Item_func :public Item_func_or_sum
     print(&str, QT_NO_DATA_EXPANSION);
     my_error(ER_DATA_OUT_OF_RANGE, MYF(0), type_name, str.c_ptr_safe());
   }
+  inline double normalize_signed_zero(double value)
+  {
+    return value == 0.0 ? 0.0 : value;  // Converts -0.0 to +0.0
+  }
   inline double raise_float_overflow()
   {
     raise_numeric_overflow("DOUBLE");
@@ -308,6 +312,7 @@ class Item_func :public Item_func_or_sum
   {
     return std::isfinite(value) ? value : raise_float_overflow();
   }
+
   /**
     Throw an error if the input BIGINT value represented by the
     (longlong value, bool unsigned flag) pair cannot be returned by the