We have monitored non-biodegradable soluble COD of leachates derived from two different landfill test cells, which were constructed at Odayeri Sanitary Landfill and operated with (C2) and without (C1) leachate recirculation for 1080 days. Refuse height and the placement area of test cells were 5 m and 1250 m(2) (25 m x 50 in), respectively. For leachates of both cells, initial inert soluble COD fraction (f) increased from it is initial value of 0.01 to around 0.1 after 300 days of operation. Due to the development of anaerobic conditions, the value showed an increasing trend and the maximum value of 0.4 was reached on day 600. Several suitable models were also fitted to the experimental data on the basis of statistical reasoning. So as to evaluate the goodness of obtained fits, the calculated values of the sum of squares due to error (SSE), R-square, the residual degrees of freedom (DFE). adjusted R-square, and root mean square errors (RMSE) associated with the model results were compared. Logistic model for C1 test cell and Gompertz model for C2 test cell gave the best fits to the experimental data. Moreover, using the fitted model parameters, pollution loads. and BOD/COD ratios in leachates from C1 (control) and C2 (recirculation) cells were estimated and deeply discussed. The results of the study can be satisfactorily used to predict change in the composition of leachate over time, which may help to obtain better effluent quality in biological treatment of leachate. (c) 2005 Elsevier B.V. All rights reserved.