Capillary Electrophoresis (CE) is a separation technique that can be used as a sample pre-treatment step in chemical analysis. When coupled with a detection technique, identification of chemical species can be performed on the basis of the elution signals. However, the sensor signals are often complicated by high signal noise, varying baseline and overlapping peaks. There is thus a need for a signal processing technique capable of robustly detecting peaks in acquired sensor data. Here, we report on an algorithm that utilises the Continuous Wavelet Transform (CWT) for the detection of analyte peaks.The algorithm that has been developed makes use of a wavelet equal to the first derivative of a Gaussian function and has been successfully applied to data obtained from a CCD sensor fabricated on a polymer microfluidic separation chip. The algorithm operates by taking the CWT of the sensor response. It then analyses patterns in the local maximum and minimum points evident across scales in the CWT coefficients to find the peaks in the time series data. The performance of two versions of the algorithm have been compared for synthetic data sets each with known baseline, peaks and noise. The improved algorithm has been shown to successfully find peaks with a high sensitivity and low False Discovery Rate within a range of sensitivities.