The content of metal elements in the root influences the growth of sorghum. Laser-Induced Breakdown Spectroscopy (LIBS) is an ideal technology for rapidly detecting metal elements in crops.In this paper, a quantitative analysis method of metal elements in sorghum roots was established based on laser-induced breakdown spectroscopy and wavelength selection algorithm based on variable dimension particle swarm optimization-combined moving window (VDPSO-CMW). We collected 27 sorghum samples with different Na and Fe concentrations under sodium salt stress. For LIBS spectra of sorghum roots, the VDPSO-CMW algorithm was used to screen the characteristic bands related to Na and Fe, and PLS quantitative analysis model was constructed. After VDPSO-CMW algorithm optimization, the determination coefficient of cross validation (R2CV) of the PLS model for Na in sorghum root was 0.962, which was 6.5% higher than that before optimization. The root means square error of cross validation (RMSECV) was 1.261, which was 37.7% lower than thatbefore optimization; the determination coefficient of prediction (R2P) was 0.988, which was 16.8% greater than that before optimization. While the root means square error of prediction (RMSEP) was 1.063, which was 72.1% lower than that before optimization. After VDPSO-CMW algorithm optimization, the R2CV of the PLS model for Fe in sorghum root was 0.956, which was 7.4% higher than that before optimization; the RMSECV was 5.095, which was 37.1% lower than that before optimization; the R2P was 0.955, which was 4.3% higher than that before optimization; while the RMSEP was 6.438, which was 27.3% lower than that before optimization. The results show that the VDPSO-CMW wavelength selection algorithm can eliminate the LIBS bands affected by self-absorption, spectral line interference, and other factors and improve the accuracy of quantitative analysis. The combination of this algorithm and LIBS technology can not only realize the rapid and accurate determination of Na and Fe in sorghum roots but may also apply to the quantitative analysis of other samples and elements.