PDFs can be very challenging, as you are probably already aware from other posts in the forum, because there are many different PDF writing programs which do not necessarily follow the 'rules' specified at the time they were written, let alone the more up to date methods in the PDF specifications. Indeed some of the things that do actually work when creating a PDF may not be in the official specification. This situation is not unique to PDF documents - many software applications have similar instances, most notably Microsoft's Internet Explorer during the early years.
If I have understood correctly your original PDF document has data presented in columns with borders. If the borders are graphical presentation into which the text of the data has been 'inserted' then the border will not appear - from what you describe that seems to be the situation you have. If the column borders were inserted as text I would expect that you should be able to see them, although it is possible that the character use may be filter in some way during the interpretation process. If you use a recent version of the Adobe PDF Reader to convert the document to text and still don't see the borders I think it would be safe to assume that they are included as graphics rather than text.
I think that leaves you with finding a way to interpret the output as it is.
Some thoughts about that:
You can use the PDF Import options '[I]scaling[/I]' value (Stretch[/I] in V10) to see if making the output wider will prevent columns on different lines overlapping. Playing with the 'Monospaced' settings may also have some useful effect but it is rather specific to the font content of the pdf and you may first need to discover whether the font(s) used were fixed width or proportional. Adobe reader should offer you the option to view the Document Properties (in the FILE menu) and the FONT tab should list all the fonts found in the document.)
You might be able to use the first attempted output as the basis for further manipulation of the format of each line to re-present the lines in a more usable form, export the result and then create a model based in that exported file.
If the best result you can get results in non-overlapping columns but with different widths you might still be able to use the MCR functionality by employing the clever idea from Olly Bond described in a recent [URL="http://www.monarchforums.com/showthread.php?t=2510#article3"]'Monarch Report'[/URL].
/LISTOf course it is more than possible that none of these ideas is of any use to you with your particular PDFs. They can have such diverse contents and internal constructions that commonly used approaches just don't quite work for individual files or, worse, work sometimes but not with reliable consistency. Working with a truly representative example file and some free thinking ideas is the only way I know of that might find a solution of the regularly used approaches, mention above, don't quite get you the result you need.