Yes, since the registration is done by intensity matching. The program does include an attempt to match intensity scales as well as positions, but it probably won't deal well with intensity mismatches of over a factor of 2 or so. You have a factor of about 10, which is the cause of the problem. Basically, I've never seen this problem before, so didn't allow for this possibility.
I could put a fix in for this, but until that is done, the only real fix is for you to manually scale the inputs to match (roughly). One way to do that would be to create a separate template for matching. For example, if you want to use sub-brick #3 from run 1 as the template, but it needs to be scaled up by a factor of 10, then
3dcalc -a 'run1+orig[3]' -expr '10*a' -prefix run1_Xtemplate
will produce a 1 sub-brick dataset that you could use as the base dataset for 3dvolreg.
bob cox