diff --git a/rtl/os2/heapsize.pas b/rtl/os2/heapsize.pas
deleted file mode 100644
index 3df5e17834..0000000000
--- a/rtl/os2/heapsize.pas
+++ /dev/null
@@ -1,25 +0,0 @@
-program heapsize;
-
-var a:longint;
-
-procedure writeheapsize;
-
-begin
-    asm
-        movl $0x7f00,%ax
-        xorl %edx,%edx
-        call ___syscall
-        mov %eax,_A
-    end;
-    writeln(a);
-end;
-
-begin
-    writeheapsize;
-    asm
-        movl $0x7f00,%ax
-        movl $327680,%edx
-        call ___syscall
-    end;
-    writeheapsize;
-end.
diff --git a/rtl/os2/mkhello.cmd b/rtl/os2/mkhello.cmd
deleted file mode 100644
index 8d77902ed0..0000000000
--- a/rtl/os2/mkhello.cmd
+++ /dev/null
@@ -1 +0,0 @@
-ld -o helloos2.exe prt0.o prt1.o helloos2.o sysos2.o dos.a emx.a dosinit.o
diff --git a/rtl/os2/o2rtlb1.pas b/rtl/os2/o2rtlb1.pas
deleted file mode 100644
index 31411f825f..0000000000
--- a/rtl/os2/o2rtlb1.pas
+++ /dev/null
@@ -1,83 +0,0 @@
-program testread;
-{uses crt;}
-var
-  cadena,cadena2 : string;
-  number : real;
-begin
-  {clrscr;}
-  cadena2 := 'Previous string';
-  write ('Enter the string ');
-  readln (cadena);
-  writeln ('You entered ',cadena);
-  writeln ('Previous string was ',cadena2);
-  write ('Enter a number ');
-  readln (number);
-  writeln ('Number entered was ',number);
-  readln;
-end.
-
-{(I have retyped now because my computer is not connected to the net, but I
-think that there are no errors).
-
-Now you can do some tests:
-
-1- Compile and run the program as is (that is, using crt). You will find that
-      a) the program does not erase the screen (that is normal because we have
-commented clrscr), but the cursor goes to the first line, thus overwriting the
-screen.
-   b) While the program is expecting the string to be entered, some of the keys
-do not work correctly: Backspace advances some spaces (just like tab), tab key
-does not work and the cursor keys write garbage. (however this is only in the
-screen, because if you have erased a part of the string it will be actually
-erased).
-   c) Once you have press return, the message 'You entered...' appears in the
-same line as the text entered.
-
-2- Uncomment the clrscr call, cokpile and execute. Point a of test 1 will be
-solved (the screen is erased, so nothing is overwritten), but points b and c
-persist.
-
-3- Comment 'uses crt' and 'clrscr'. Now you will not be using crt. Now:
-   a) Point a of test 1 does not appear: the program begins to write in the
-next line, it does not overwrite anything.
-   b) Now all the keys (tab, backspace..) work as expected.
-   c) Now the message 'You entered...' appears in the following line, so point
-c of test 1 is also solved.
-   d) BUT it writes only 'You entered', WITHOUT writing the string cadena (!).
-It writes also 'Previous string was previous string', so the problem is in
-readln and not in writeln.
-
-4- To see if the problem is only in the string vars, uncomment the definition
-of number, and also the three lines at the end that deal with number. Now ld
-gives the following error message:
-
-testread.pp:0 (testread.o): undefined symbol READ_TEXT_INTEGER referenced from
-text segment.
-
-This error happens with 'uses crt' and also without it.
-
-5- Define number as word. Regardless of crt we get the following error from ld:
-
-testread.pp:0 (testread.o): undefined symbol READ_TEXT_WORD referenced from
-text segment.
-
-6- Uncomment 'uses crt' if it was commented, and change the definition of
-number as real. The program will compile, and it will print the number,
-although in the same line as the input.
-
-7- Finally, comment 'uses crt' again. This time it will also compile and link,
-but it gives a runtime error!
-
-Laufzeitfehler 106 bei 66422
-
-This error is shown before printing the number.
-
-I expect that these bug report will be useful to debug the RTL. Tonight I will
-try to work in the blockwrite problem.
-
-Best regards
-
-Ramon
-
--- 
-}