Abstract: 294 seed samples of 9 soybean varieties in Hulunbeier were collected and identified in the spring of 2008. Taking GB/T3543.5-1995 as the standard, we designed the field arrangement by 4N principle, adopted artificial precise dibble seeding and high quality chemical weeding, recorded the main characteristics of the varieties and then conducted the comparison. The results showed that the average emergence rate was 87.0%, 92.2% samples had the rate higher than 80%, 168 plots had hard seedlings, which accounted for 57.1% of the total, but there were not many plots with high hard seedlings, and the hard seedling did not seriously affect the seed quality. The purity identification of samples from 259 plots reached the standard, the total rate was 88.1%. 68 plots were high purity sample plot, no mixed strain was found, accounted for 23.1% of the total; the sample purity of 10 plots was less than 90% , accounted for 3.4%, and there were 4 plots which had problem samples (different varieties), accounted for 1.4%. The emergence rate and purity were excellent, but the quality of a few low purity samples was unacceptable. The purity identification showed that there were aggregations of high and low purity samples, the samples from high quality enterprise were excellent, and the samples from inferior enterprises had low quality. The experiment confirms that the field plot planting is stable and reliable, and can still be one of the main methods for seed authenticity and purity identification.