The correct analysis and interpretation of intracardiac electric records in the presence of supraventricular arrhythmias highly depends on the precise detection of activation times. However, state-of-art automatic detectors show limitations, making manual revision of the detections mandatory. In this work we propose a novel automatic detector based on wavelet analysis and curve-length transformation. A database consisting in 60 manually annotated intracardiac bipolar records, 15 during atrial fibrillation (AF) and 45 during atypical atrial flutter (AAFl) was used to test the proposed algorithm. Results obtained were compared to results obtained with two classic automatic detectors (Botteron and Teager-Kaiser energy operator based detector) in terms of sensibility and positive predictive value. The proposed algorithm showed better performance than Botteron’s both in AF and AAFl situations (SENS %: 98.5 v s 95.4 in AAFl / 91.21 vs 84.76 in AF; PPV %: 98.47 vs 96.32 in AAFl / 91.25 vs 86.11 in AF )and similar to those obtained with Teager-Kaiser operator based detector (SENS %: 98.5 vs 98.62 in AAFl / 91.21 vs 91.06 in AF; PPV %: 98.47 vs 96.32 in AAFl / 91.25 vs 91.05 in AF ). These results suggest that the proposed strategy may provide a more robust behaviour than classic algorithms, although further efforts must be put in refining the algorithm and it also must be validated with a bigger database.