The miniaturization of GSM handsets creates nonlinear acous-tical echoes between microphones and loudspeakers when the signal level is high (hands-free communication). A comparison of several nonlinear echo cancellation methods have been studied in a recent paper [1]. In these paper, we propose to studdy the effects of echo cancellation on automatic speech recognition performance. We use the speech recognition system (SPEERAL) developed by the LIA laboratory. The quality of the echo cancelers is also evaluated using the standard ERLE (echo return loss enhancement) measure.