Background: Intravenous thrombolysis and intra-arterial thrombectomy are proven emergency treatments for acute ischaemic stroke, but they require rapid delivery to selected patients within specialist services. National audit data have shown that treatment provision is suboptimal. Objectives: The aims were to (1) determine the content, clinical effectiveness and day 90 cost-effectiveness of an enhanced paramedic assessment designed to facilitate thrombolysis delivery in hospital and (2) model thrombectomy service configuration options with optimal activity and cost-effectiveness informed by expert and public views. Design: A mixed-methods approach was employed between 2014 and 2019. Systematic reviews examined enhanced paramedic roles and thrombectomy effectiveness. Professional and service user groups developed a thrombolysis-focused Paramedic Acute Stroke Treatment Assessment, which was evaluated in a pragmatic multicentre cluster randomised controlled trial and parallel process evaluation. Clinicians, patients, carers and the public were surveyed regarding thrombectomy service configuration. A decision tree was constructed from published data to estimate thrombectomy eligibility of the UK stroke population. A matching discrete-event simulation predicted patient benefits and financial consequences from increasing the number of centres. Setting: The paramedic assessment trial was hosted by three regional ambulance services (in north-east England, north-west England and Wales) serving 15 hospitals. Participants: A total of 103 health-care representatives and 20 public representatives assisted in the development of the paramedic assessment. The trial enrolled 1214 stroke patients within 4 hours of symptom onset. Thrombectomy service provision was informed by a Delphi exercise with 64 stroke specialists and neuroradiologists, and surveys of 147 patients and 105 public respondents. Interventions: The paramedic assessment comprised additional pre-hospital information collection, structured hospital handover, practical assistance up to 15 minutes post handover, a pre-departure care checklist and clinician feedback. Main outcome measures: The primary outcome was the proportion of patients receiving thrombolysis. Secondary outcomes included day 90 health (poor status was a modified Rankin Scale score of > 2). Economic outputs reported the number of cases treated and cost-effectiveness using quality-adjusted life-years and Great British pounds. Data sources: National registry data from the Sentinel Stroke National Audit Programme and the Scottish Stroke Care Audit were used. Review methods: Systematic searches of electronic bibliographies were used to identify relevant literature. Study inclusion and data extraction processes were described using Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Results: The paramedic assessment trial found a clinically important but statistically non-significant reduction in thrombolysis among intervention patients, compared with standard care patients [197/500 (39.4%) vs. 319/714 (44.7%), respectively] (adjusted odds ratio 0.81, 95% confidence interval 0.61 to 1.08; p = 0.15). The rate of poor health outcomes was not significantly different, but was lower in the intervention group than in the standard care group [313/489 (64.0%) vs. 461/690 (66.8%), respectively] (adjusted odds ratio 0.86, 95% confidence interval 0.60 to 1.2; p = 0.39). There was no difference in the quality-adjusted life-years gained between the groups (0.005, 95% confidence interval –0.004 to 0.015), but total costs were significantly lower for patients in the intervention group than for those in the standard care group (–£1086, 95% confidence interval –£2236 to –£13). It has been estimated that, in the UK, 10,140–11,530 patients per year (i.e. 12% of stroke admissions) are eligible for thrombectomy. Meta-analysis of published data confirmed that thrombectomy-treated patients were significantly more likely to be functionally independent than patients receiving standard care (odds ratio 2.39, 95% confidence interval 1.88 to 3.04; n = 1841). Expert consensus and most public survey respondents favoured selective secondary transfer for accessing thrombectomy at regional neuroscience centres. The discrete-event simulation model suggested that six new English centres might generate 190 quality-adjusted life-years (95% confidence interval –6 to 399 quality-adjusted life-years) and a saving of £1,864,000 per year (95% confidence interval –£1,204,000 to £5,017,000 saving per year). The total mean thrombectomy cost up to 72 hours was £12,440, mostly attributable to the consumables. There was no significant cost difference between direct admission and secondary transfer (mean difference –£368, 95% confidence interval –£1016 to £279; p = 0.26). Limitations: Evidence for paramedic assessment fidelity was limited and group allocation could not be masked. Thrombectomy surveys represented respondent views only. Simulation models assumed that populations were consistent with published meta-analyses, included limited parameters reflecting underlying data sets and did not consider the capital costs of setting up new services. Conclusions: Paramedic assessment did not increase the proportion of patients receiving thrombolysis, but outcomes were consistent with improved cost-effectiveness at day 90, possibly reflecting better informed treatment decisions and/or adherence to clinical guidelines. However, the health difference was non-significant, small and short term. Approximately 12% of stroke patients are suitable for thrombectomy and widespread provision is likely to generate health and resource gains. Clinician and public views support secondary transfer to access treatment.