As salinity is an important indicator in marine geology, ecology, breeding, and other fields, accurate, rapid, and continuous measurement of salinity is of great significance in marine investigations. At present, the seawater salinity detection methods used in practice are mainly based on the principle that the conductivity and refractive index parameters of the water change with the concentration of elements, which are composed of salinity change. However, these methods quantitatively analyze salinity values by measuring other parameters (electrical or optical parameters) that may change depending on the salinity of the water, rather than the mass fraction of the components that make up the salinity. Moreover, when the salinity value of seawater water changes substantially or the proportion of various common components composing salinity changes significantly, the detection accuracy of the above methods is difficult to guarantee. Therefore, a spectral approach, LIBS, and the Raman spectroscopy combination method for salinity analyzation, LRSS, were proposed to provide a new option. The main idea of this approach is to use the two spectral detection methods, LIBS and Raman, to determine the mole values of cations and non-monatomic anions in per unit quality (1 kg) of water, respectively. Then the mole value of the chloride ion, which is the main monatomic anion in seawater, can be determined according to the electrically neutral principle. Based on all the obtained molar values and the molar mass of each ion, the salinity of the water sample can be determined. To demonstrate the performance of this new method, we compared it with LIBS under laboratory conditions and found that, when non-monatomic anions are present in the water, the accuracy of LRSS is significantly improved compared to using the LIBS method alone. Moreover, we also compared the LRSS with the other two traditional methods through the 11 water samples configured and found that the absolute value relative error of the LRSS is only 2.63% when the salinity and components concentration change is in the possible range, which is better than the conductivity method 0.53 times and better than the refractive index method 1.52 times.