In this paper, a new approach to the optimal retrieval of the ocean color based on the variational method is developed by setting up a rational target functional combining with the Broydor Fletcher, Goldfarb, Shanno (BFGS) optimal algorithm. The numerical tests and the exemplary retrievals are carried out and compared with the statistical retrievals and the optimal retrievals based on the genetic algorithm. The results show that this approach enjoys a higher accuracy as compared to the statistical method and a higher efficiency as compared to the genetic algorithm. The optimal retrieval method presented in this paper provides a new idea for the ocean color inversion and could also be used as a reference for the direct assimilation of the satellite data into the ecological models. In this paper, a new approach to the optimal retrieval of the ocean color based on the variational method is developed by setting up a rational target functional combining with the Broydor Fletcher, Goldfarb, Shanno (BFGS) optimal algorithm. The numerical tests and the example retrievals are carried out and compared with the statistical retrievals and the optimal retrievals based on the genetic algorithm. The results show that this approach enjoys a higher accuracy as compared to the statistical method and a higher efficiency as compared to the genetic algorithm. The optimal retrieval method presented in this paper provides a new idea for the ocean color inversion and could also be used as a reference for the direct assimilation of the satellite data into the ecological models.