A 59-year-old male patient without the history of heart disease was referred for coronary angiography due to typical chest pain and signs of myocardial ischemia during exercise stress test. On admission, the patient was in sinus rhythm. After successful PCI, the patient was enrolled in a study to investigate esophageal long-term electrocardiography (eECG) comprising simultaneous rhythm monitoring with surface and esophageal leads during 16 h. The surface ECG (sECG) was recorded using a two-channel Holter ECG recorder with a sampling frequency of 1,024 Hz allowing continuous rhythm monitoring for 7 days (Lifecard CF, Spacelabs Healthcare, USA). The eECG was recorded using an esophageal ECG electrode (Esosoft 6S, FIAB, Italy) connected to a dedicated two-channel eECG recorder—a proprietary development—with a sampling frequency of 512 Hz. We registered two bipolar channels (interelectrode spacings of 60 and 15 mm). The eECG was well tolerated; the patient reported only a slight pharyngeal foreign body feeling without any impact on his common daily activities such as eating, talking and sleeping. The sECG was analyzed using an automatic beat classifier system (Pathfinder, Spacelabs Healthcare, USA). The software correctly detected five atrial premature beats (APB’s) but no other atrial arrhythmias. In the eECG, we unmasked another 40 APB’s, as well as 178 episodes with atrial ectopic rhythm [mean duration ± SD 25 ± 31 beats (Fig. 1)]. This report illustrates the limited sensitivity of surface long-term ECG for atrial rhythm disorders. For the sECG, software-based automatic classification of atrial arrhythmias is accomplished by calculating the ratio of the R–Rinterval to be classified and the preceding R–R-interval(s). This quotient is then compared to a coupling interval threshold, e.g., the Pathfinder software uses by default 66 % of the preceding R–R-interval duration. The number of detected atrial arrhythmias strongly depends on this threshold: atrial ectopic beats are only identifiable if they are premature enough as compared to this threshold. The sECG uses R–R-intervals as surrogate markers of the true atrial activity, while morphologic features of atrial signals are not considered. The lack of a direct classifier for atrial arrhythmias is a major limitation of sECG monitoring in cases without evident change in R–R cycle duration. Moreover, reliable detection of atrial arrhythmias is of increasing importance for risk stratification, because it has been shown that excessive supraventricular ectopic activity [1] as well as atrial premature beats [2–4] are associated with an increased risk of stroke and atrial fibrillation. Esophageal electrocardiography may provide an elegant way out by offering excellent atrial signals that additionally allow waveform analysis for the direct and reliable classification of atrial rhythm. Lead movements due to eating or respiration can confound esophageal electrocardiography. We exclude that the morphologic changes are caused by such artifacts for the ClinicalTrials.gov Identifier: NCT01436344.