The fitting software QMXRS (Quantitative analysis of Micro-Energy Dispersive X-ray fluorescence spectra) which is used for fitting micro-energy X-ray fluorescence spectrum focused by poly-capillary optics is developed in our Lab with Python language. The signal de-noising of the wavelet transform, background subtraction, energy calibration and characteristic peak fitting of spectrum is carried out by QMXRS. The ploy-capillary X-ray optics used here can improve the intensity of X-ray and change the distributions of X-ray fluorescence spectra. The background prediction model of QMXRS can effectively retain peak information and improve the accuracy of data processing of energy spectra. On the other hand, the constraint of the full width at half maximum(FWHM) of the Gaussian peak is added to the conventional nonlinear least-squares fitting. The Micro-Energy dispersive X-ray fluorescence spectra of the NIST 610 standard sample is used to demonstrate the performances of QMXRS. The results show that our QMXRS is better than PyMca and QXAS.