Inductively coupled plasma mass spectrometry was used for determination of U in 175 samples of drinking water from central and northern parts of Moravia in the Czech Republic. The paper is devoted to validation of the method and optimization of conditions for the determination. Limits of detection and determination were found, the suitability of calibration dependence was investigated, accuracy and precision of determination were checked and also the influence of sample storage in laboratory and signal stability were examined. For estimation of the mean U value in all samples above the determination limit, the median 1.09 0.36 µg.l^-1 was used due to the rejected assumption of normal distribution. In almost 20 % of samples, the U concentration was below the detection limit. The U levels higher than 5 µg.l^-1 were found in 11 % of samples.