>I have forms which must be translated programmatically into postscript files. Labels may be word-wrapped. I know you can use TXTWIDTH and FONTMETRIC to get the pixel length of a string.
>
>Can anyone tell me how to apply this to determine where the breaks might be in a label, given the width and height of the label?

You may start with a loop, adding content of the label word by word, and then calculating the width after each word, until the length of the visual representation grows more than the label length.

I'm not sure the PostScript rendering engine will yield the same results as what you get from TxtWidth, though. You may be off by one word here and there.