Poor insight in schizophrenia spectrum disorders (SSD) is linked with negative outcomes. This single-centre, assessor-blind, parallel-group 1-year follow-up randomised controlled trial (RCT) tested whether metacognitive training (MCT) (compared to psychoeducation) may improve insight and outcomes in outpatients with SSD assessed: at baseline (T0); after treatment (T1) and at 1-year follow-up (T2). Insight (primary outcome) was measured with (i) the Schedule for Assessment of Insight-Expanded version- (SAI-E), including illness recognition (IR), symptom relabelling (SR), treatment compliance (TC) and total insight scores (TIS); and (ii) the Beck Cognitive Insight Scale (BCIS). Between-group comparisons were nonsignificant, while within the MCT group (but not within controls) there was a significant medium effect size for improved TIS at T2 (d = 0.67, P = 0.02). Secondary outcomes included cognitive measures: Jumping to Conclusions (JTC), Theory of Mind (ToM), plus symptom severity and functioning. Compared to psychoeducation, MCT improved the PANSS excitement (d = 1.21, P = 0.01) and depressed (d = 0.76, P = 0.05) factors at T2; and a JTC task both at T1 (P = 0.016) and at T2 (P = 0.031). Participants in this RCT receiving MCT showed improved insight at 1-year follow-up, which was associated with better mood and reduced JTC cognitive bias. In this pilot study, no significant benefits on insight of MCT over psychoeducation were detected, which may have been due to insufficient power.