Background: Smoking prevalence among people with psychosis remains high. Providing Very Brief Advice (VBA) comprising: i) ASK, identifying a patient's smoking status ii) ADVISE, advising on the best way to stop and iii) ACT (OFFER), offering a referral to specialist smoking cessation support, increases quit attempts in the general population. We assessed whether system-level changes in a UK mental health organisation improved the recording of the provision of ASK, ADVISE, ACT (OFFER) and consent to referral to specialist smoking cessation support (ACT (CONSENT)). Methods: We conducted a study using a regression discontinuity design in four psychiatric hospitals with patients who received treatment from an inpatient psychosis service over 52 months (May 2012-September 2016). The system-level changes to facilitate the provision of VBA comprised: A) financially incentivising recording smoking status and offer of support (ASK and ACT (OFFER)); B) introduction of a comprehensive smoke-free policy; C) enhancements to the patient electronic healthcare record (EHCR) which included C1) a temporary form to record the financial incentivisation of ASK and ACT (OFFER) C2) amendments to how VBA was recorded in the EHCR and C3) the integration of a new electronic national referral system in the EHCR. The recording of ASK, ADVISE, ACT (OFFER/CONSENT) were extracted using a de-identified psychiatric case register. Results: There were 8976 admissions of 5434 unique individuals during the study period. Following A) financial incentive, the odds of recording ASK increased (OR: 1.56, 95%CI: 1.24-1.95). Following B) comprehensive smoke-free policy, the odds of recording ADVICE increased (OR: 3.36, 95%CI: 1.39-8.13). Following C1) temporary recording form, the odds of recording ASK (OR:1.99, 95%CI:1.59-2.48) and recording ACT (OFFER) increased (OR: 4.22, 95%CI: 2.51-7.12). Following C3) electronic referral system, the odds of recording ASK (OR:1.79, 95%CI: 1.31-2.43) and ACT (OFFER; OR: 1.09, 95%CI: 0.59-1.99) increased. There was no change in recording VBA outcomes following C2) amendments to VBA recording.