Continuous background spectrum is an important influencing factor of laser induced breakdown spectrometry (LIBS) quantitative measurement of multi-element heavy metals in water. According to the characters that the background spectrum changes slowly and full width at half maximum of characteristic spectral lines is narrow, the sliding window integral slope algorithm for removing continuous backgrounds is studied. Seting appropriate window width and slope threshold to identify the background spectrum and characteristic peaks, the assignment is carried out on window background, the background spectrum is extracted completely. The experimental results show that the signal-to-background ratios of plumbum and copper are 5.7 times and 1.95 times respectively to that before removing the background, the relative standard deviations are reduced by 2% and 2.5% respectively. The sliding window integral slope algorithm is a stable and effective method for removing backgrounds in LIBS, and also provids an reference for removing backgrounds in other optical spectrum analysis technology.