To investigate the genetic variation in perilla germplasm resources across different regions of Guizhou and to facilitate the identification of superior traits and the breeding of high-quality varieties， this study analysed and evaluated 15 major quantitative traits in 348 perilla germplasm resources using methods such as genetic diversity analysis， correlation analysis， principal component analysis， cluster analysis and comprehensive scoring. The results showed that the coefficients of variation for the 15 traits ranged from 4.44% to 56.70%. Among these， the coefficients of variation for seed weight per plant and total number of ears per plant were relatively high， at 56.70% and 53.30%， respectively. The genetic diversity indices ranged from 0.99 to 2.01， with the highest indices observed for seed weight per plant， plant height and number of main stem nodes， at 2.01， 1.85 and 1.77， respectively. Correlation analysis revealed 52 pairs of significant or highly significant correlations among the 15 traits. In particular， seed weight per plant was significantly positively correlated with plant height， stem diameter， number of main stem nodes， total number of ears per plant and stearic acid content， and was highly significantly positively correlated with oil content and growth stage. Principal component analysis identified six principal components， with contribution rates of 19.14%， 14.03%， 11.10%， 9.24%， 8.53% and 7.46%， respectively， accounting for a cumulative total of 69.50%. Cluster analysis divided the 348 perilla germplasm resources into four groups： Group I comprises short-stalked， thick-stemmed plants with numerous spikes and a long main spike， and high linolenic acid content； Group II comprises plants with numerous branches， early maturity， relatively high oil content， and a balanced fatty acid composition； Group III comprises early-maturing plants with notably high protein and oleic acid content； Group IV comprises tall-stemmed， high-yielding， late-maturing， with high oil and stearic acid content. The 13 unique germplasm resources selected from these four groups provide a material foundation for the precise and efficient breeding of perilla and the in-depth exploration of superior genes.Based on an evaluation using the composite F-score， 15 perilla germplasm resources with an F-score of ≥1 and excellent overall traits were selected； these may serve as a reference for future perilla breeding.