Technical Abstract: CQESTR simulates the effect of several management practices on soil organic carbon stocks. The model had been calibrated and validated in temperate regions. Our objective was to calibrate the CQESTR model and evaluate its performance for simulating carbon dynamics as affected by tillage practices in two tropical soils (Ultisol and Oxisol) in southeastern and northeastern Brazil. In the southeast (20.75 S 42.81 W) tillage systems consisted of no-tillage (NT), reduced tillage (RT) (one disc plough and one harrow leveling (RT1), or one heavy disc harrow and one harrow leveling (RT2)) and conventional tillage (CT) (two heavy disc harrow followed by one disc plough and two harrow leveling). In the northeast (7.55 S 45.23 W), tillage systems consisted of NT, RT (one chisel plow and one harrow leveling) and CT (one disk plow, two heavy disk-harrowing and two harrow leveling). With respect to initial conditions (native forest) CQESTR estimated a decrease in total organic carbon (TOC) stocks in plowed and no-tillage systems. In 2006, in the southeast, TOC stocks were 28.8, 23.7, 23.2 and 22.0 Mg ha-1 under NT, RT2, RT1 and CT, respectively and in the northeast, were 36.0, 33.8 and 32.5 Mg ha-1 under NT, RT and CT, respectively. The model estimated carbon emissions varying from 1.28 (NT) to 3.67 Mg ha-1 year-1 (CT) in the southeast and from 1.13 (NT) to 3.04 (CT) Mg ha-1 year-1 in the northeast. CQESTR underestimated TOC stocks in both sites, especially under no-tillage systems, indicating that adjustments (e.g. the inclusion of clay mineralogy) are necessary for more accurate simulation of TOC in the tropics. In spite of this, measured and simulated values of TOC in the 0-20 cm layer were well correlated (northeast, R2=0.88, p<0.05; southeast, R2 =0.94, p<0.01) showing acceptable performance in the prediction of TOC in two tropical soils of Brazil.