Background: Echocardiographic assessment of systolic and diastolic function during atrial fibrillation (AF) is challenging. This study evaluates the prognostic value of strain in patients with AF and suggests a novel approach on how to take into account the varying heart cycle lengths in AF. Methods: Echocardiograms from 204 patients with AF during examination were analyzed offline. Patients with known heart failure (HF) were excluded. Peak global longitudinal strain (GLS) was averaged from 18 myocardial segments. To adjust for the varying heart cycle lengths, we indexed GLS with the square root of the RR-interval, (GLS/√(RR)). The composite endpoint included incident HF, stroke, myocardial infarction and all-cause mortality. Results: During a median follow-up of 2.4 years, 82 patients (40%) reached the composite endpoint. Decreasing GLS/√(RR) was significantly associated with the composite endpoint, and the risk of reaching the endpoint increased significantly per 1%/sec^1/2 decrease in strain (HR 1.13, 95% CI 1.07–1.20, p < 0.001). GLS/√(RR) remained an independent predictor even after adjustment for various risk factors and conventional echocardiography (LVEF and E/e′) (HR 1.10, 95% CI: 1.02–1.19, p = 0.017). In contrast, GLS did not remain a significant predictor after adjusting for the same variables (p = 0.07), neither did LVEF (p = 0.11). Conclusion: Decreasing GLS/√(RR) was significantly associated with increased risk of an adverse outcome and remained an independent predictor after multivariable adjustment. Indexing GLS with the square root of the RR-interval can counteract the variable cycle length in AF patients and GLS/√(RR) offers a more convincing risk-stratification assessment in AF patients compared with GLS.