The risk stratification in patients presenting with acute dyspnoea remains a challenge. We therefore conducted a prospective, observational cohort study enrolling 292 patients presenting to the emergency department with acute dyspnoea. A proteomic approach for antibody-free targeted protein quantification based on high-end MS was used to measure LTBP2 [latent TGF (transforming growth factor)-binding protein 2] levels. Final diagnosis and death during follow-up were adjudicated blinded to LTBP2 levels. AHF (acute heart failure) was the final diagnosis in 54% of patients. In both AHF ( P <0.001) and non-AHF ( P =0.015) patients, LTBP2 levels at presentation were significantly higher in non-survivors compared with survivors with differences on median levels being 2.2- and 1.5-fold respectively. When assessing the cause of death, LTBP2 levels were significantly higher in patients dying from pulmonary causes ( P =0.0005). Overall, LTBP2 powerfully predicted early pulmonary death {AUC (area under the curve), 0.95 [95% CI (confidence interval), 0.91–0.98]}. In ROC (receiver operating characteristic) curve analyses for the prediction of 1-year mortality LTBP2 achieved an AUC of 0.77 (95% CI, 0.71–0.84); comparable with the predictive potential of NT-proBNP [N-terminal pro-B-type natriuruetic peptide; 0.77 (95% CI, 0.72–0.82)]. Importantly, the predictive potential of LTBP2 persisted in patients with AHF as the cause of dypnea (AUC 0.78) and was independent of renal dysfunction (AUC 0.77). In a multivariate Cox regression analysis, LTBP2 was the strongest independent predictor of death [HR (hazard ratio), 3.76 (95% CI, 2.13–6.64); P <0.0001]. In conclusion, plasma levels of LTBP2 present a novel and powerful predictor of all-cause mortality, and particularly pulmonary death. Cause-specific prediction of death would enable targeted prevention, e.g. with pre-emptive antibiotic therapy.