Although many animal studies24- 37(Table 2) support the importance of oxidative mechanisms in neurologic disorders, potential publication bias may favor positive studies, a problem less likely to occur with human studies. Randomized controlled clinical trials of compounds with antioxidant properties have yielded positive, negative, marginal, or conflicting results, both in neurologic and nonneurologic disorders. Compounds tested in neurologic disease include vitamin E, tirilazad, N-acetylcysteine, ebselen, selegiline, idebenone, and extract of Gingko biloba. Diseases studied include Parkinson disease, Alzheimer disease, multi-infarct dementia, amyotrophic lateral sclerosis (ALS), Huntington disease, acute ischemic stroke, subarachnoid hemorrhage, head and spinal cord injury, and intractable childhood epilepsy. Unfortunately, none of the clinical trials performed to date has measured markers of oxidative injury as a surrogate marker of drug efficacy, either in pretrial dose-finding studies or in subgroups of patients during the trials. This is important because some of the compounds used may not be effective antioxidants in vivo in humans, or alternatively the compounds may be effective but have been given in suboptimal doses. Some compounds (eg, vitamin E, probucol) may have antioxidant or pro-oxidant effects depending on dosage, concomitant treatments, and study models.38,39 These considerations underscore the importance of pretrial dose-finding studies using novel markers of free radical injury now available, such as isoprostanes,40,41 8-hydroxy-2-deoxyguanosine,42 or 3-nitrotyrosine,43 reliable markers of lipid, DNA, and protein oxidation, respectively. Measurement of potential reduction of biochemical indices of free radical injury should be made in subgroups of patients as a surrogate marker of antioxidant effectiveness in future controlled trials.44 The objective measurement of antioxidant effectiveness could also provide critical information in individual treatment trials of putative antioxidants in rare neurologic disorders in which large prospective studies are not possible.