Internal contamination by compounds of cobalt radioisotopes occurs time to time at nuclear power plants. Intakes and committed effective doses are estimated by biokinetic models described in ICRP publications. The paper deals with a case of internal contamination of a worker engaged in a maintenance task at NPP Dukovany. In this case significant discrepancy was observed between intakes based on various datasets (whole body counting, analysis of urine and faeces) when default model setting was used. The reason of this phenomenon was searched for. Three different least square methods of fits were used to find out possible effect of a fitting method. The measured data were fitted by set of biokinetic functions, which covered all intake ways (ingestion and inhalation) and types (M, S, different AMADs and different f1) of the contaminant. The biokinetic model of cobalt needs further improvements as to find better agreement between data fit from direct measurements and bioassay.