There are many fresh jujube varieties. The different variety has different quality. In addition, dehiscent fruit easily rot and the rotten fruit contaminated the full fruit very rapidly. It is necessary to discriminate the jujube varieties and dehiscent fruit to reduce the storage loss. The objective is to discriminate varieties and dehiscent Fruit of fresh jujube using near infrared (NIR) spectroscopy. Two jujube varieties were investigated. Smoothing, multiplicative scatter correction, the first derivative and second derivative methods were adopted to pretreat the spectra. The regression coefficient and principal component analysis were used to select wavenumber. Multilayer perceptron artificial neural network was used to build varieties and dehiscent fruit qualitative discrimination model. The results showed that the varieties and dehiscent fruit could be correctly discriminated and both the discrimination accuracy rates were 100%. Hence, near infrared spectroscopy could achieve to identify the variety of Dongzao and Lizao, and dehiscent fruit and intact fruit.