Especially when writing PHP scripts for use on different servers, it is a very good idea to explicitly set the internal encoding somewhere on top of every document served, e.g.

mb_internal_encoding("UTF-8");

This, in combination with mysql-statement "SET NAMES 'utf8'", will save a lot of debugging trouble. 

Also, use the multi-byte string functions instead of the ones you may be used to, e.g. mb_strlen() instead of strlen(), etc.

all together 

<?php
// ------------------------------------------------------------ 

header('Content-Type: text/html; charset=UTF-8');

mb_internal_encoding('UTF-8'); 
mb_http_output('UTF-8'); 
mb_http_input('UTF-8'); 
mb_regex_encoding('UTF-8'); 

// ------------------------------------------------------------ 
?>

Be aware that the strings in your source files must match the encoding you specify by mb_internal_encoding.  It appears the Parser loads raw bytes from the file and refers to its internal encoding to determine their actual encoding.

To demonstrate, the following outputs as espected when the /source/ file is Latin-1 encoded:

<?php
    mb_internal_encoding("iso-8859-1");
    mb_http_output( "UTF-8" );
    ob_start("mb_output_handler");

    echo "???<br/>";

    ?>???

Now, a typical use of mb_internal_encoding is shown as follows.  Make the change to "utf-8" but leave the /source/ file encoding unchanged:

<?php
    mb_internal_encoding("UTF-8");
    mb_http_output( "UTF-8" );
    ob_start("mb_output_handler");

    echo "???<br/>";

    ?>???

The output will just show the <br/> tag and no text.

Save the file as UTF-8 encoding and then the results will be as expected.

Note that mb_internal_encoding is not necessary in PHP 5.6

In response to mortoray at ecircle-ag dot com:

The characters display fine as long as you set the Encoding to something more "Latin 1" compatible (i.e. US-ACSII, ISO-8859-1, ISO-8859-1, or  Windows 1252). PHP.net auto-detects to UTF-8