A rapid method for determining magnesium in a multivitamin pharmaceutical preparation has been validated. This element is analysed by flame atomic absorption spectrometry after dissolution of the sample in acid medium. Linearity, precision, accuracy and robustness of the method have been determined, and detection and quantification limits have been calculated. Linearity of response was verified for concentrations ranging from 0.05 to 0.40 mg x l(-1) of magnesium. Correlation coefficient of the calibration straight lines was always > or =0.9999. Repeatability of the method gave relative standard deviation (R.S.D.) of 0.6%. Reproducibility of the method calculated after analysis of samples by the same analyst in different days (day-to-day fluctuation) and by two different analysts in different days (analyst-to-analyst fluctuation) gave relative standard deviation of 1.1 and 1.6%, respectively. Mean recoveries of magnesium obtained after spiking sample placebos with increasing amounts of magnesium chloride ranged from 98.9 to 100.8%. Robustness of the method evaluated by changing different experimental conditions under which analyses were performed, gave relative standard deviation from 0.2 to 0.5%. Limits of detection and quantification were 3.8 and 7.0 microg of Mg per gram of sample, respectively. Results show the suitability of the method for direct measurement of magnesium in a water-soluble multivitamin pharmaceutical preparation.