Serum CBZ concentration was determined by gas liquid chromatographic method in 14 epileptic patients (32 samples) treated initially with CBZ alone, and the following results were obtained: 1. The relationship between serum CBZ concentration and CBZ dose was expressed by a regression line that sloped sharply upwards at first: high serum concentration was obtained from small dose up to the end of three weeks after initiation of CBZ administration. After the 3rd--4th treatment week, regression coefficient was reduced by half, and the regression line had a gentler slope. 2. As the treatment was continued longer, serum CBZ concentration and serum concentration/dose ratio became lower. The relationship between duration of CBZ treatment and serum concentration/dose ratio was expressed by a regression curve representing a quadratic function with negative coefficient, the ratio being high in early stage of the treatment. This regression curve shows that the ratio reached its maximum value at the 10.5 days after initiation of the treatment. The ratio was stabilized at a low level about 3--4 weeks after that. 3. From the results, serum CBZ concentration was confirmed to be more dependent on duration of the treatment than on dose. This was inferred, on the basis of documentary research, to be attributable to the autoinduction of CBZ metabolism, the inference of which seemed to be supported by the results of measurement of time-course change in serum concentration in five cases of the present series. The results in one of the said five cases suggest that serum CBZ concentration varies greatly within 12 hr. 4. On the basis of these results, it is proposed that spontaneous decrease in serum CBZ concentration must be countered not merely by increase in CBZ dose, but by increase in the number of dosages/day for suppression of excessive diurnal change in serum concentration as well.