Causal inference using area-level socioeconomic measures is challenging due to risks of residual confounding and imprecise specification of the neighbourhood-level social exposure. By using multi-linked longitudinal data to address these common limitations, our study aimed to identify protective effects of neighbourhood socioeconomic improvement on premature mortality risk. We used data from the Canadian Community Health Survey, linked to health administrative data, including longitudinal residential history. Individuals aged 25-69, living in low-socioeconomic status (SES) areas at survey date (n = 8335), were followed up for neighbourhood socioeconomic improvement within 5 years. We captured premature mortality (death before age 75) until 2016. We estimated protective effects of neighbourhood socioeconomic improvement exposures using Cox proportional hazards models. Stabilized inverse probability of treatment weights (IPTW) were used to account for confounding by baseline health, social and behavioural characteristics. Separate analyses were carried out for three exposure specifications: any improvement, improvement by residential mobility (i.e. movers) or improvement in place (non-movers). Overall, 36.9% of the study cohort experienced neighbourhood socioeconomic improvement either by residential mobility or improvement in place. There were noted differences in baseline health status, demographics and individual SES between exposure groups. IPTW survival models showed a modest protective effect on premature mortality risk of socioeconomic improvement overall (HR = 0.86; 95% CI 0.63, 1.18). Effects were stronger for improvement in place (HR = 0.67; 95% CI 0.48, 0.93) than for improvement by residential mobility (HR = 1.07, 95% 0.67, 1.51). Our study provides robust evidence that specific neighbourhood socioeconomic improvement exposures are important for determining mortality risks.