Is there a better way to count # of records in a table than doing count(*)?

Sometimes it takes forever and consumes lots of resources to count records in a huge table, say with more than 200+ millions records.