Late functional improvement between 3 and 12 months poststroke occurs in about one in four patients with ischaemic stroke, more commonly in lacunar strokes. It is unknown whether this late improvement is associated with better long-term clinical or health economic outcomes.In a prospective, population-based cohort of 1-year ischaemic stroke survivors (Oxford Vascular Study; 2002-2014), we examined changes in functional status (modified Rankin Scale (mRS), Rivermead Mobility Index (RMI), Barthel Index (BI)) from 3 to 12 months poststroke. We used Cox regressions adjusted for age, sex, 3-month disability and stroke subtype (lacunar vs non-lacunar) to examine the association of late improvement (by ≥1 mRS grades, ≥1 RMI points and/or ≥2 BI points between 3 and 12 months) with 5-year mortality and institutionalisation. We used similarly adjusted generalised linear models to examine association with 5-year healthcare/social-care costs.Among 1288 one-year survivors, 1135 (88.1%) had 3-month mRS >0, of whom 319 (28.1%) demonstrated late functional improvement between 3 and 12 months poststroke. Late improvers had lower 5-year mortality (aHR per mRS=0.68, 95% CI 0.51 to 0.91, p=0.009), institutionalisation (aHR 0.48, 0.33 to 0.72, p<0.001) and healthcare/social care costs (margin US$17 524, -24 763 to -10 284, p<0.001). These associations remained on excluding patients with recurrent strokes during follow-up (eg, 5-year mortality/institutionalisation: aHR 0.59, 0.44 to 0.79, p<0.001) and on examining late improvement per RMI and/or BI (eg, 5-year mortality/institutionalisation with RMI/BI: aHR 0.73, 0.58 to 0.92, p=0.008).Late functional improvement poststroke is associated with lower 5-year mortality, institutionalisation rates and healthcare/social care costs. These findings should motivate patients and clinicians to maximise late recovery in routine practice, and to consider extending access to proven rehabilitative therapies during the first year poststroke.