BACKGROUND: Plant stem structural characteristics are crucial factors determining plant lodging resistance, while high throughput methods for rapid surveys of these traits are still lacking in sorghum.RESULTS: Among 103 sorghum accessions, two kinds of stem powders (dry and water-washed) were subject to visible and near-infrared spectra acquisition, and 16 models (combinations) for stem structural characteristics were generated, revealing that the support vector machine regression model has significant positive effects on the prediction of stem structural characteristics while powder type and pretreatment of spectra has minor effects on the prediction of stem structural characteristics. In addition, we found that stem structure characteristics were positively correlated with agronomic traits but negatively correlated with lodging index which is the criterion that negatively accounts for plant lodging resistance.CONCLUSION: This study for the first time provided a precise and high throughput method for the prediction of sorghum stem structural characteristics based on spectra, which could facilitate the improvement of lodging resistance in crop breeding.