The control of binocular rivalry (BR) was reintroduced as an example of selective attention after several decades. Ss passively observed BR in a stereoscopic tachistoscope and, at an undetermined phase of rivalry, received an instruction signal to attend to one of the rivalry patterns. Letter test stimuli were presented to either the instructed (I) eye or noninstructed (NI) eye 1, 3, 5, or 7 sec following the instruction signal. The difference between the eyes in recognition performance (I-NI) provided an objective measure of control, which was correlated with previously used measures based on subjective reports of rivalry alternation rates. Subjective and objective measures showed correlated increases as a result of practice of control. The fact that the objective measure reached a peak value at the 3-sec delay of test stimuli suggested a minimum time to shift attention and maximum time to maintain attention in BR. In addition, it was found that the control of BR results in a change of signal strength with no associated change in the use of confidence ratings, a result similar to that found with selective attention in dichotic listening.