Spontaneous breathing trials (SBTs) are among the most commonly employed techniques to facilitate weaning from mechanical ventilation. The preferred SBT technique, however, is still unclear. To clarify the preferable SBT (T-piece or pressure support ventilation [PSV]), we conducted this systematic review. We then searched the MEDLINE, EMBASE, SciELO, Google Scholar, CINAHL, ClinicalTrials.gov, and Cochrane CENTRAL databases through June 2015, without language restrictions. We included randomized controlled trials involving adult subjects being weaned from mechanical ventilation comparing T-piece with PSV and reporting (1) weaning failure, (2) re-intubation rate, (3) ICU mortality, or (4) weaning duration. Anticipating clinical heterogeneity among the included studies, we compared prespecified subgroups: (1) simple, difficult, or prolonged weaning and (2) subjects with COPD. We summarized the quality of evidence for intervention effects using the GRADE (Grading of Recommendations Assessment, Development, and Evaluation) methodology. We identified 3,674 potentially relevant studies and reviewed 23 papers in full. Twelve studies (2,161 subjects) met our inclusion criteria. Overall, the evidence was of very low to low quality. SBT technique did not influence weaning success (risk ratio 1.23 [0.94-1.61]), ICU mortality (risk ratio 1.11 [0.80-1.54]), or re-intubation rate (risk ratio 1.21 [0.90-1.63]). Prespecified subgroup analysis suggested that PSV might be superior to T-piece with regard to weaning success for simple-to-wean subjects (risk ratio 1.44 [1.11-1.86]). For the prolonged-weaning subgroup, however, T-piece was associated with a shorter weaning duration (weighted mean difference -3.08 [-5.24 to -0.92] d). In conclusion, low-quality evidence is available concerning this topic. PSV may be associated with lower weaning failure rates in the simple-to-wean subgroup. In contrast, in prolonged-weaning subjects, T-piece may be related to a shorter weaning duration, although this is at high risk of bias. Further study of the difficult-to-wean and COPD subgroups is required.