During thoracic impedance signal acquisition, noise is inherently introduced and hence, denoising is required to allow for accurate event detection. This paper investigates the effectiveness of Ensemble Emperical Mode Decomposition to filter random noise. The performance of the EEMD method is compared with an optimal FIR filter and wavelet denoising. The IMF selection for signal reconstruction in the EEMD denoising method is optimized using a sequential search. Denoising performance was evaluated by the SNR and the accuracy in event detection after filtering. When all criteria are taken into account, wavelet seems to outperform both EEMD and FIR denoising.