We estimate short- and long-term causal impacts of a type-2 diabetes mellitus (T2DM) diagnosis on lifestyle behaviours. We employ a fuzzy regression discontinuity design exploiting the exogenous cut-off value in the diagnosis of T2DM provided by a biomarker (glycated haemoglobin, HbA1c). We make use of unique administrative longitudinal data from Spain and focus on the impact of a diagnosis on clinically measured BMI, smoking and alcohol consumption. We find that, following a T2DM diagnosis, individuals appear to reduce their weight in the short-term. These effects are particularly large among obese individuals and those diagnosed with depression. Patients who are younger, still in the labour market and healthier also present increased short-term probabilities of quitting smoking. In addition, we provide evidence of statistically significant long-term impacts of a T2DM diagnosis on BMI up to three years from the diagnosis. Our results are consistent across parametric and non-parametric estimations with varying bandwidths. Overall, our findings suggest the relevance of health information in affecting changes in key lifestyle behaviours.