In this research, a modified split-plot design model (SPDM) is introduced, where the mean part of the SPDM is tailored with a three-parameter Gompertz function, which modifies the SPDM to an intrinsically nonlinear SPDM. The model is applied to a balanced 2-replicated 3×42 mixed-level split-plot design experiment to determine the irrigation effect on four varieties of rice trial at four different rates of nitrogen fertilizer. The SPDM parameters were estimated through the methods of estimated generalized least squares (EGLS) and restricted maximum likelihood estimation (REML) for estimating the SPDM variance components. The parameter estimates are compared to estimates from ordinary least squares (OLS) and maximum likelihood estimation (MLE) estimates of the variance using four median adequacy measures and three information criteria for goodness of fit. The results obtained shows that the modified SPDM is a good fit, and its EGLS-REML estimates are of better reliability and adequacy compared to the OLS and EGLS-MLE techniques.