The number of automatic mode switch (AMS) has been used to measure the efficacy of atrial pacing in limiting atrial fibrillation (AF). We investigated the impact of length and contiguity on the specificity of AMS in detecting AF episodes in 24 recipients of dual chamber pacemakers with sick sinus syndrome and paroxysmal AF. An AMS algorithm and intracardiac electrogram recordings (IEGM) were activated in all patients to distinguish true arrhythmic events from unnecessary AMS. The length of AMS and the contiguity, that is, the probability of occurrence of another AMS within 5 minutes before or after AMS were examined to increase the specificity of the AMS. During a mean follow-up of 5 +/- 3 months, 250 AMS were collected. The IEGM analysis confirmed a true AF episode in 193 of 250 AMS (77.2%). Using the contiguity criterion, 47 of 57 (82.5%) inappropriate AMS episodes were isolated (there were no other AMS within 5 minutes), whereas 54 of 193 (27.9%) appropriate AMS episode were isolated. Adopting both length and contiguity criteria the specificity of AMS in detecting true AF episodes increased from 77.2% to 93.2% at the cost of 11.9% loss of original sensitivity. Combining the length and contiguity criteria, we were able to improve the specificity of the AMS in the detection of AF.