I have both scientific and ethical concerns about the study by Harris et al1 on intercessory prayer. First, the study was not randomized. Patients were systematically assigned to study groups by a method that the authors expected to be random in net effect; that is not the same thing. The approach used by Harris et al more often results in accidental unblinding and systematic bias from unexpected systematic behavior than does true randomization. Although this is relatively unlikely to have caused problems in this specific study, the term randomized should be reserved for studies that are actually randomized.