<p>Research on sustainment of implemented evidence-based practices (EBPs) for people with psychosis has been limited in mental health services. This prospective cohort study extended a cluster-randomized trial (trial registration number NCT03271242) of implementation of four EBPs in Norwegian mental health services. In the trial 39 clinical units were randomized to receive implementation support for one of two chosen EBPs. This study aimed to measure sustainment for two biological EBPs (physical health care, antipsychotic medication management) and two psychosocial EBPs (family psychoeducation, illness management and recovery), and to explore factors influencing sustainment. Fidelity to the EBP with implementation support was measured after 18 months of implementation support and at 36 months (18 months post-implementation), additionally key informants reported factors influencing sustainment. Among 27 sites with high or moderate fidelity at 18 months, 20 (74%) sustained the practice at the same or higher level at 36 months; 10 of 13 sites, most with psychosocial EBPs, sustained high fidelity, and 10 of 14 sites, most with biological EBPs, sustained moderate fidelity. Sustaining sites demonstrated significantly greater efforts to continue the EBP, while non-sustaining sites encountered greater barriers including leadership change or key clinician turnover. With efforts to continue the EBP, most sites sustained the practice with high fidelity for psychosocial EBPs and moderate fidelity for biological practices. A clinical director or champion determined to continue the practice together with a trained clinical staff were the two main ingredients for sustainment.</p>