> Given that I am worried about differences BETWEEN conditions, is there a better way to control for RT?

Do you mean that the average RT across trials is different among the conditions? Do you want to compare the conditions while controlling RT at some common value? If so, read the following part from the 3dDeconvolve -help:

*** [12 Jul 2012] You can now specify the value to subtract from
each modulation parameter -- this value will replace the
subtraction of the average parameter value that usually happens.
To do this, add an extra parameter after the option, as in
-stim_times_AM2 1 timesAM.1D 'BLOCK(2,1)' :5.2:x:2.0
The extra argument must start with the colon ':' character, and
there should be as many different values (separated by ':') as
there are parameters in the timing file (timesAM.1D above).
==> In the example above, ':5.2:x:2.0' means
subtract 5.2 from each value of the first parameter in timesAM.1D
subtract the MEAN from each value of the second parameter
(since 'x' doesn't translate to a number)
subtract 2.0 from each value of the third parameter
==> What is this option for, anyway? The purpose is to facilitate
GROUP analysis the results from a collection of subjects, where
you want to treat each subject's analysis exactly the same
way -- and thus, the subtraction value for a parameter (e.g.,
reaction time) should then be the mean over all the reaction
times from all trials in all subjects.

Gang