Actually translation, not transcription, star at the Start codon and stops at a Stop codon. And they are three possible lecture frames ( http://en.wikipedia.org/wiki/Reading_frame ) so it's possible that you find multiple stop codon but it might that they are derived from different lecture frames.

although stop codons represent the end of translation, it is totally possible and correct that you might find several stop codons, since it also depends on the ORF (Open Reading Frame) (http://en.wikipedia.org/wiki/Open_reading_frame).

This basically means that you have to take codons three by three, so there are six ways of grouping codons for each sequence (three for one strand, three for the complementary strand). If you are programatically looking at this, you should take into accout this when scanning through the sequence, otherwise you might find stop codons that do not really stop the process.

You can take a look also at this website: http://www.ncbi.nlm.nih.gov/gorf/gorf.html which shows you the open reading frames for any sequence you search.