Bipolar disorders of types I and II, even when treated by currently standard options, show a marked excess of depressive morbidity. Treated, type I patients in mid-course or from the onset of illness are ill, overall, 50 % of weeks of follow-up, and 75 % of that unresolved morbidity is depressive. Currently widely held impressions are that bipolar depression typically is poorly responsive to antidepressants, that treatment-resistant depression (TRD) is characteristic of the disorder, and that risk of mania with antidepressant treatment is very high. However, none of these views is supported consistently by available research. TRD may be more prevalent in bipolar than unipolar mood disorders. Relatively intense research attention is directed toward characteristics and treatments of TRD in unipolar depression, but studies of bipolar TRD are uncommon. We found only five controlled trials, plus 10 uncontrolled trials, providing data on a total of 13 drug treatments, all of which involved one or two trials, in 87 % as add-ons to complex, uncontrolled regimens. In two controlled trials, ketamine was superior to placebo but it is short-acting and not orally active; pramipexole was weakly superior to placebo in one controlled trial; three other drugs failed to outperform controls. Other pharmacotherapies are inadequately evaluated and nonpharmacological options are virtually untested in bipolar TRD. The available research supports the view that antidepressants may be effective in bipolar depression provided that currently agitated patients are excluded, that risk of mania with antidepressants is only moderately greater than risk of spontaneous mania, and that bipolar TRD is not necessarily resistant to all treatments.