Introduction Up to 10% of patients presenting with chest pain and elevated troponin levels demonstrate non-obstructive coronary arteries on angiography, posing a clinical challenge in diagnosis, prognosis and management. The final diagnosis has important implications for the patient, including prescription for treatment and fitness for permissible activities, occupation and medical insurance. Cardiovascular magnetic resonance (CMR) is superior to other cardiac imaging modalities in tissue characterisation. We hypothesised that CMR, when performed early using conventional and novel tissue characterisation techniques, can determine the cause of acute myocardial injury in these patients and provide a diagnosis. Methods One hundred and twenty (n = 120) patients (mean age 50 ± 17 yrs; 50% female) presenting with chest pain, positive troponin I (normal <0.04, median 3.99, range 0.07 to >60 μg/L) and non-obstructive coronaries were prospectively recruited.Early CMR at 1.5T (median 3 days, IQR 1–6 days) included cine, T2W (dark-blood STIR), T1-mapping (ShMOLLI) and LGE imaging. Findings were compared to 50 controls matched for age and gender distributions.Image analysis included: the detection of oedema comparing T2 signal intensity of myocardium to skeletal muscle (>2.0) or remote myocardium (>2 SD); myocardial T1 times (areas of injury defined as an area of ≥40 mm2 with T1 >990 ms as validated for detecting oedema); and presence of LGE. Results When CMR was performed early using only conventional techniques (cine, T2W and LGE), there was a high diagnostic yield of 95%. Oedema was detected in 79% and LGE in 61% of patients.Based on CMR findings, including the type, pattern and regional distribution of injury, the commonest diagnosis was myocarditis (37%), followed by Takotsubo cardiomyopathy (23%), myocardial infarction (18%), acute regional stunning (9%; wall motion abnormality with oedema but no LGE), dilated cardiomyopathy (4%), hypertrophic cardiomyopathy (3%), and missed pulmonary embolism (1%). In 11/21 (52%) of patients with MI, a patent foramen ovale (PFO) was demonstrated on echocardiography with agitated saline contrast, suggesting these patients may have suffered a paradoxical coronary embolism. The remaining 5.0% (n = 6) of patients had no findings on T2W and LGE imaging. However, T1 mapping localised areas of injury in 4 out of the remaining 6 patients, improving the detection rate to 98%. Conclusions Early CMR using conventional and novel T1-mapping techniques has a high diagnostic yield in patients presenting with acute chest pain, positive troponins but non-obstructive coronaries. T1 mapping detected additional areas of abnormality when conventional CMR was “normal”, improving the detection rate to 98%. Early multiparametric CMR is able to localise areas of affected myocardium and is useful in the further management or diagnostic workup in this patient cohort.