Your question doesn't tell what repeat are you looking into nor what is the source sequence. Is it TEs or is it simple repeats. I have some experience with the former.
In case this is the human or some other well studied model organism, pre-annotated repeat information might (must) be available from RepeatMasker or UCSC Genome Browser site.
In case your organism is not covered, check what repeat is in question. For repeats like TEs, they have canonical lengths. This info. can be obtained from GIRI (after free reg.). Probably you have it already.
I have used RepeatMasker and it adjudges whether a given sequence is of repeat origin or not based upon its own thresholds (you can of course use a sensitive flag), but that is it. You need not bother more, like whether alignment length was sufficient or not.
In case you are checking transcribed sequences, or genomic units like exons or gene locus, which can have fraction of the genomic repeat element present, 'length' can be a consideration.
In case of exons, to assign it as repeat harbouring, I used threshold of 10% of exon length or >25bp.
Lastly, things might be different if you are looking into simple repeats. My experience is with TEs