In this study, we propose a parameter estimation method for genome-scale dynamic flux balance (DFBA) models. A bilevel optimization problem is reformulated as a differential-algebraic equation (DAE) optimization problem and solved sequentially, using gradient-based optimization with direct sensitivity equations. The resulting solution method is computationally efficient for today's largest genome-scale metabolic models. The parameter estimation method combined with parameter selection algorithm is applied on simulated and experimental data. This paper presents the parameter estimation and selection method and numerical results of estimation of kinetic parameters of the DFBA model of anaerobic batch fermentation. The results show improved computational performance over previous approaches, thus making parameter estimation available for genome-scale DFBA models.