It’s very important to accurately predict crop root biomass for evaluation of carbon source/sink function in crop ecosystem. With the late rice of 2012, in Yujiang County, Jiangxi Province, the paper collected leaf area index (LAI) data in 4 different rice growth stages (tillering, heading, filling and full ripeness stage) and the above and below ground biomass data in full ripeness stage from 26 monitoring plots, explored the correlations between LAIs and rice root biomass, compared the LAI approach to the root shoot ratio method in the prediction of rice root biomass, and aimed to establish the prediction models of rice root biomass with multi-stage LAIs. The results showed that there was a strong correlation (r = 0.775, P＜0.01) between LAI and root biomass in rice filling stage; the prediction precision of root biomass with LAI approach was higher than that with root shoot ratio method; multi-stage LAIs could improve the prediction precision, esp. the LAIs combination in heading, filling, and full ripeness stages could build the robust prediction model of crop root biomass while the best prediction model with single stage LAI had lower prediction precision. In conclusion, the rice root biomass can be precisely predicted with LAI approach, which may make a foundation for direct prediction of crop root biomass with multi-stage LAIs in the future.