Abstract Background Secondary mitral regurgitation (SMR) is associated with increased morbidity and mortality. In these patients, left ventricular global longitudinal strain (LVGLS) and left atrial reservoir strain (LARS) have been proposed to improve risk stratification. However their prognostic value was shown to be inconsistent when considering the SMR etiology (i.e. atrial or ventricular), possibly due to be the influence of atrial fibrillation (AF) on these measures and specifically LARS. Moreover, data focusing on moderate SMR is lacking. Aim To investigate the prognostic value of LVGLS and LARS in patients with SMR stratified according to etiology, and to identify optimal threshold for both measures. Methods Patients (n=1061; age 68±11years , 59% male) with moderate SMR were retrospectively included and classified as ASMR or VSMR, and further subdivided into "sinus rhythm (SR) subgroup" (ASMR-SR: n=68 , VSMR-SR: n=435) or as "AF subgroup" (ASMR-AF: n=284 , VSMR-AF: n=274), if patients had history of AF. The endpoint was a composite of heart failure events and all-cause mortality. Threshold values for LVGLS and LARS were defined based on spline curve analysis. Results As compared to VSMR, ASMR patients were more frequently female (61% vs. 35%, p<0.001), had less comorbidities and were less symptomatic (i.e. NYHA functional class III-IV in 11% of ASMR vs. 24% of VSMR, p<0.001). For the echocardiographic characteristics, VSMR patients showed lower LV ejection fraction and LVGLS, as compared to ASMR patients. For both ASMR and VSMR, patients in the AF subgroups had larger LA volume index, lower LARS and more severe tricuspid regurgitation (Figure 1). During a median follow-up of 78 (IQR: 44-111) months, 27% of ASMR-SR, 37% of ASMR-AF, 50% of VSMR-SR and 72% of VSMR-AF group reached the combined endpoint. Separated multivariable cox regression analyses were performed in ASMR and VSMR patients (Table 1): for both etiologies, LARS and LVGLS were independently associated with the endpoint after correcting for other significant variables at the univariate analysis, including the SR or AF group and the interventions during follow-up. In addition, a stepwise addition of LVGLS and LARS to a basic model led to a significant difference in the likelihood ratio for both groups. The thresholds of LVGLS and LARS associated with an increased risk of the combined endpoint differed per etiology. For LVGLS the threshold was <16% for ASMR, and <12% for VSMR. When considering LARS assessment, different thresholds were identified according to the SR or AF group: <18% in ASMR-SR; <15% in ASMR-AF; <17% in VFMR-SR and <12% in VFMR-AF. Conclusion In moderate SMR patients, LVGLS and LARS are independently associated with outcomes both in ASMR and VSMR and could be considered to improve risk stratification. However, lower threshold for both parameters should be considered based on etiology and on the presence of AF.