The use of the ECG for atrial fibrillation (AF) in clinical daily practice is still limited to its diagnosis. Recent research shows however that ECG-derived parameters can also be used to assess the spatiotemporal properties of AF. Specifically, the complexity of the f-waves in the ECG reflects the complexity of the fibrillatory conduction during AF and therefore can be used for quantification of the degree of electrophysiological alterations in the atria. This information might be useful for guiding AF therapy and might form the basis for classification of AF. This review focuses on technical and mathematical aspects of ECG-based atrial complexity assessment and its potential ability to guide treatment strategies.