Desulfurizations of a model oil (hexadecane containing dibenzothiophene (DBT)) and a diesel oil by immobilized DBT-desulfurizing bacterial strains, Gordona sp. CYKS1 and Nocardia sp. CYKS2, were carried out. Celite bead was used as a biosupport for cell immobilization. Seven-eight cycles of repeated-batch desulfurization were conducted for each strain. Each batch reaction was carried out for 24 h. In the case of model oil treatment with strain CYKS1, about 4.0 mM of DBT in hexadecane (0.13 g sulfur l(oil)(-1)) was desulfurized during the first batch, while 0.25 g sulfur l(oil)(-1) during the final eighth batch. The mean desulfurization rate increased from 0.24 for the first batch to 0.48 mg sulfur l(dispersion)(-1) h(-1) for the final batch. The sulfur content in the light gas oil was decreased from 3 to 2.1 g l(oil)(-1) by strain CYKS1 in the first batch. The mean desulfurization rate was 1.81 mg sulfur l(dispersion)(-1) h(-1), which decreased slightly when the batch reaction was repealed. No significant changes in desulfurization rate were observed with strain CYKS2 when the batch reaction was repeated. When the immobilized cells were stored at 4 degrees C in 0.1 M phosphate buffer (pH 7.0) for 10 days, the residual desulfurization activity was about 50 similar to 70% of the initial value. (C) 2000 Published by Elsevier Science B.V. All rights reserved.