Do you own Application Vantage? If not you can use another sniffer program like sniffer.
Take an AV trace of yourself with a browser running through the transaction, then take an av trace of qaload running the transaction. Compare the results. You may have extra requests, or just a caching issue. Check it out first.
I believe that this is most often seen for subrequests generated the likes of java script in which case QALoad hare-codes the subrequests (depending on settings). You should try it with QALoad 5.6 utilizing the java script execution level or playing with the HTML/HTTP modes, if at all possible.
Are you using the c scripts or .vistree? When you view the sript do you see the subrequests hard-coded there? If so, you may need to manually manipulate the script to match with actual browser/app. behavior.
If not, well then... I have seen behavior like this and have wondered about it. I have captureing manual vs. QALoad through a capture proxy like Fiddler (see this post for links), and I usually see more subrequests comming from QALoad than when captured manually. I've never quite understood it myself and I haven't run enough tests with QALoad to see a diff. yet.
A problem is a difference between what is perceived and what is desired, that
we want to reduce (Dewey 1933)