Abstract Background The cognitive side-effects of ECT are minimized by individualized, supra-threshold dosing with brief pulse electrical stimuli. Unilateral ECT is associated with fewer cognitive sequelae but bilateral ECT is possibly more effective. Little is known of the relative effectiveness and tolerability of the two placements in the treatment of elderly, severely depressed inpatients. Methods Patients of five public aged psychiatry services and a private psychiatric hospital who received right unilateral ECT ( n = 47) dosed on average at 3 times seizure threshold or bitemporal ECT ( n = 16) dosed at 1.5 times threshold completed an abbreviated Rey Auditory Verbal Learning Test, a visual memory test and a section of the Autobiographical Memory Inventory as close as possible to 24 h after the first or second treatment and again after the fifth or sixth treatment. This design was intended to maximize recruitment of severely depressed patients with a limited ability to consent and cooperate with testing. Results Only 35% of eligible patients completed both assessments, mostly due to refusal or lack of capacity to consent. Moderate dose unilateral and bilateral ECT produced equivalent improvements in mood. There was a tendency for scores on most cognitive tests to decline more with bilateral than unilateral ECT but these differences were statistically significant only for immediate verbal memory and autobiographical memory. Conclusions Our findings suggest that bilateral ECT is no more effective as an antidepressant than moderately dosed unilateral ECT, at least on a short-term basis, and confers a slightly greater risk of cognitive impairment. This supports the rationale of prescribing unilateral ECT in the first instance in this vulnerable clinical population. Limitations It proved impossible to recruit most ECT recipients, limiting the capacity to generalize findings to all aged patients. Reports concerning ECT should list recruitment rates to help set findings in context.