Will be helpful if you can provide precise, step-by-step instructions
for reproducing the problem.

Here it is (for 3.5.1) on my czech locale machine with windows-1250 
default encoding:

1) Start ide, create a new project, mount some empty directory

2) Go to Tools/Options,Editing/Java sources, go to Strings table 
property and change USER to a name with accented characters (like 
USER=Jeíek ). Close the options.

3) Create a new class using the Java Class template. Give it Somename 
and press Finish. This will produce correct result, which may look 
like (note the accented characters in localized date and in @author):

/*
 * Somename.java
 *
 * Created on 18. záøí 2003, 22:46
 */

/**
 *
 * @author  Jeíek
 */
public class Somename {
    
    /** Creates a new instance of Somename */
    public Somename() {
    }
    
}

4) Now go to Tools/Options,Editing/Java Sources,Expert pane, and 
change Default Encoding property to UTF-8.
Close options.

5) Repeat step 3) creating class Somename2. This time, it will look 
like this (note the damaged accented characters caused by generating 
the source code using platform default windows-1250 encoding instead 
of UTF-8, and then using it like if it was UTF-8):

/*
 * Somename2.java
 *
 * Created on 18. z��� 2003, 22:49
 */

/**
 *
 * @author  Je��ek
 */
public class Somename2 {
    
    /** Creates a new instance of Somename2 */
    public Somename2() {
    }
    
}

The #65533; shown here show up as squares in the editor

*** This issue has been marked as a duplicate of 55104 ***