In three experiments, listeners identified speech processed into narrow bands and presented to the right ("target") ear. The ability of listeners to ignore (or even use) conflicting contralateral stimulation was examined by presenting various maskers to the target ear ("ipsilateral") and nontarget ear ("contralateral"). Theoretically, an absence of contralateral interference would imply selectively attending to only the target ear; the presence of interference from the contralateral stimulus would imply that listeners were unable to treat the stimuli at the two ears independently; and improved performance in the presence of informative contralateral stimulation would imply that listeners can process the signals at both ears and keep them separate rather than combining them. Experiments showed evidence of the ability to selectively process (or respond to) only the target ear in some, but not all, conditions. No evidence was found for improved performance due to contralateral stimulation. The pattern of interference found across experiments supports an interaction of stimulus-based factors (auditory grouping) and task-based factors (demand for processing resources) and suggests that listeners may not always be able to listen to the "better" ear even when it would be beneficial to do so.