The aim of this study was to investigate soil C and N stocks and soil physical properties under a cotton (Gossypium hirsutum) based system using no‐till (NT) and conventional tillage (CT). The experiment was established as a randomized block design with four replications of six treatments as follows: [NT1], NT system with crop rotation of cotton/soybean (Glycine max)/maize (Zea mays) + brachiaria (Brachiaria ruziziensis); [NT2], NT system with crop rotation of soybean/maize + brachiaria/cotton; [NT3], NT system with crop rotation of maize + brachiaria/cotton/soybean; [CR1], CT with annual rotation (soybean/cotton); and [CR2], CT with biannual rotation (soybean/soybean/cotton) and [CT1] CT with cotton monoculture. An area of native Cerrado vegetation was used as the reference or baseline. Soil samples were obtained from 0‐ to 5‐, 5‐ to 10‐, 10‐ to 20‐, 20‐ to 30‐, 30‐ to 60‐, and 60‐ to 100‐cm depths. After 9 years of cultivation, soil under NT systems contained higher C concentrations than those under CT (p < 0.05); however, significant differences were restricted to the 0‐ to 5‐cm layer. Stocks of C and N in the 0‐ to 30‐cm depth ranged from 36.2 to 39.8 Mg ha−1 and from 2.00 to 2.22 Mg ha−1, respectively. Despite higher C concentration in the upper layer, NT treatments (NT1, NT2, and NT3) had soil C stocks similar to those of CR1 and CT1. In comparison with Cerrado, C stocks in soil under all cultivated treatments were reduced by 27% in the top 1‐m depth, which represents a C loss of 35 Mg C ha−1 upon conversion from the Cerrado to cultivated land. The least soil bulk density among all treatments was observed for NT1. However, there were no differences in macroporosity and microporosity among NT and CT systems even after 9 years of cultivation. In comparison with CT, principal benefits of NT systems for C sequestration and soil physical improvements are restricted to the surface layer.