Antipsychotic drugs are the mainstay of treatment for schizophrenia but they have little effect on core negative symptoms or cognitive impairment. To meet the deficiencies of current treatments, novel potential compounds are emerging from preclinical research but translation to clinical success has been poor. This article evaluates the possibility that cognitive and physiological abnormalities in schizophrenia can be used as central nervous system biomarkers to predict, in healthy volunteers, the likely efficacy of entirely new pharmacological approaches to treatment. Early detection of efficacy would focus resource on rapidly developing, effective drugs. We review the relevance of selected cognitive and physiological abnormalities as biomarkers in schizophrenia and three of its surrogate populations: (i) healthy volunteers with high trait schizotypy; (ii) unaffected relatives of patients; and (iii) healthy volunteers in a state of cortical glutamate disinhibition induced by low-dose ketamine. Several biomarkers are abnormal in these groups and in some instances there has been exploratory work to determine their sensitivity to drug action. They are generally insensitive to current antipsychotics and therefore their predictive validity cannot be established until novel, therapeutically useful drugs are discovered. Until then such biomarker studies can provide evidence of drugs engaging with the mechanism of interest and encouragement of the concept.