Neuroactive steroids, including testosterone, dehydroepiandrosterone (DHEA) and its sulfate (DHEA-S) might play an important role in the pathophysiology of schizophrenia. Therefore, we performed a systematic review and meta-analysis of studies comparing the levels of testosterone, DHEA and DHEA-S in patients with schizophrenia and healthy controls. We searched electronic databases from their inception until Oct 29, 2017. Effect size (ES) estimates were calculated as Hedges’ g. Data analysis was performed using random-effects models. Our analysis included 34 eligible studies, representing 1742 patients and 1604 controls. Main analysis revealed elevated DHEA-S levels in the whole group of patients (ES = 0.75, 95%CI: 0.23–1.28, p = 0.005). In subgroup analyses, patients with first-episode psychosis (FEP) had significantly higher levels of free testosterone (ES = 1.21, 95%CI: 0.30–2.12, p = 0.009) and DHEA-S (ES = 1.19, 95%CI: 0.66–1.71, p < 0.001). Acutely relapsed schizophrenia patients presented significantly higher levels of total testosterone (ES = 0.50, 95%CI: 0.21–0.70, p < 0.001). Total testosterone levels were also elevated in stable multi-episode schizophrenia (sMES) females (ES = 0.56, 95%CI: 0.33–0.80, p < 0.001) and reduced in sMES males (ES = −0.62, 95%CI: −1.07 to 0.18, p = 0.006). Increased levels of biologically active, free testosterone and DHEA-S in FEP suggest that these alterations might appear as a response to stress that becomes blunted during subsequent exacerbations of schizophrenia. Differential changes in total testosterone levels in male and female sMES patients might represent medication effects related to prolactin-releasing effects of antipsychotics.