By combining root auger and image analysis methods for root sampling and measurement, this study investigated the dynamic characteristics of cotton plant root growth and the spatial distribution of root length, diameter and area. The experiment was conducted during 2001 and 2002 at an experiment station located in Anyang city, Henan Province. A root performance experiment of different genotypes was conducted in 2001 using 4 cotton cultivars: middle season hybrid `CRI29' with Bt, middle season variety `CRI35' without Bt, middle season variety `CRI32' with Bt and early season variety `CRI37' with Bt. Plant density was 60 000 plants per hectare. In 2002, the American middle season variety `33B' with Bt was included and measured in 40 cm×40 cm×100 cm deep soil pits. The soil was loam. Climate data from a nearby weather station were used to calculate the cotton physiological development time (PDT). Computation of PDT also integrated thermal effectiveness, photoperiod effectiveness, genetic effectiveness and its interactions. Root samples were collected at four distances from the cotton row: 0, 15, 25 and 35 cm. Root samples were collected from 40 cm×40 cm×100 cm deep soil pits and sampled every 10 cm from 0 to 100 cm soil depth at each location. The roots were sampled using a root auger that had a volume of 316.73 cm 3. The roots were washed form the soil core with all cotton roots collected in the samples. Cleaned roots were scanned to a black and white image at a resolution of 200 dpi. DT-SCAN software was used to calculate root length, root diameter and root surface area. Compared to the ruler measurement method to estimate root length density (RLD), the image analysis method using DT-SCAN for estimating RLD gave similar results (R 2=0.899, n=1 318). The results indicated that the RLD of cotton averaged 1.21-1.27 mm·cm -3 during flowering and boll stage, 1.04-1.12 mm·cm -3 after boll opening, and 0.76 mm·cm -3 during the harvest period. Root diameter was significantly different among genotypes (p=0.022), with the thickest root diameter of 0.52 mm in an insect resistant hybrid cultivar and the thinnest root diameter of 0.36 mm in the early maturation cultivar. Root diameters were not significantly different between soil layers but were significantly thinner with increasing distance from the cotton row. Because of the significant differences observed in root length and diameter among soil layers, distance from plant and genotypes, we developed a root area index (RAI) that has similar biological meaning as the leaf area index. The RAI and LAI were exponentially related (R 2 = 0.799). The dynamic time course of RAI fit a logistic growth pattern (R 2=0.849) with physiological development time (PDT) when the PDT was smaller than 40, and exhibited a linear decreasing pattern (R 2=0.570-0.895) when PDT was greater than 40. The highest RAI calculated was in the insect-resistant hybrid cultivar and the lowest in early maturation cultivar. The vertical and temporal changes in RAI were as follows: RAI was the greatest in the topsoil layer (0-30 cm) before flowering, in the mid soil layer (40-60 cm) during flowering and boll stage, and in the deep layer (70-100 cm) and at locations away from the cotton row after boll opening.