Determining the effectiveness of candidate treatments for preventing hemorrhagic strokes caused by cerebral amyloid angiopathy will require clinical drug trials. This article explores tvo potential outcome markers for such trials: (1) clinical recurrence of hemorrhagic stroke, and (2) the appearance of small, clinically silent hemorrhagic lesions on gradient-echo MRI. Using pilot data from our cohort of survivors of lobar hemorrhage, we estimated the sample sizes required to demonstrate efficacy with each of these outcome markers as the study endpoint. A study with recurrent hemorrhagic stroke as the endpoint was estimated to require 145 patients per treatment arm to demonstrate a 50% reduction in the recurrence rate over a 24-month follow-up period, while a study using new hemorrhagic lesions on MRI was estimated to require 70 patients per arm for a 17-month follow-up interval. The required sample sizes could be further reduced (to 105 and 52 patients, respectively) by limiting the analysis to those at highest risk of recurrence, defined according to apolipoprotein E genotype or the presence of more than one hemorrhagic lesion at study entry. This analysis suggests that radiographic detection of small hemorrhages may be an efficient surrogate endpoint for pilot trials of promising therapeutic approaches to cerebral amyloid angiopathy.