Cognitive dysfunction is a hallmark feature of schizophrenia and is evident across all phases of the illness. While prior meta-analyses have elucidated the level and pattern of cognitive deficits in the premorbid and post-onset periods of psychosis, no meta-analyses of studies of the putative prodromal period have been published. Our primary aim is to provide a meta-analysis of neurocognitive findings from 14 studies of psychosis risk syndrome (PRS) individuals published through February 2011, and compare the resulting profile with that synthesized by meta-analyses from other periods of the disorder. Meta-analysis of 1215 PRS individuals with a mean age of 19.2 (± 3.3) and 851 healthy control subjects yielded small-to-medium impairments across nine of 10 neurocognitive domains (Cohen's d = -0.26 to -0.67). Seven studies reported on PRS individuals who later developed psychosis (n = 175) and their baseline performance level generally yielded moderate-to-large ESs (d = -0.35 to -0.84). Mild cognitive deficits are reliably and broadly present in PRS individuals, falling at a level that is intermediate between healthy individuals and those diagnosed with schizophrenia, and at a level that is comparable to those at familial ("genetic") risk and with premorbid data. Moreover, baseline neurocognition in PRS individuals who converted to psychosis showed more severe deficits than non-converters in nearly all domains. However, considerable heterogeneity of ESs across studies in many domains underscores variability in phenotypic expression and/or measurement sensitivity, and a critical need for improved reporting of sample characteristics to support moderator variable analyses.