We developed a new method for simultaneous determination of 89Sr and 90Sr with an emphasis on detectability. The samples were digested, and Sr was chemically purified followed by a single count on a liquid scintillation counter in three windows overlapping the 90Sr, 89Sr, and 90Y peaks. Gamma spectrometry was used to measure 85Sr, added for chemical recovery. The method was tested on 18 water samples spiked at levels from 9 to 242 Bq of 89Sr and 90Sr, with either single radionuclides or their mixtures. In addition, eight method blanks were measured. The data were analyzed numerically by solving a system of linear equations for 89Sr and 90Sr activities as analytes and 90Y activity as a participating component. The total uncertainties of the results were calculated numerically using variances and covariances. The average bias from the known activities was -0.3% (range from -3.6 to 3.1%) for 90Sr and - 1.5% (range from -10.1 to 5.1%) for 89Sr. The En-scores were within -1.0 and 1.0 at 95% confidence level. The detection capabilities of this method were determined by means of the decision threshold LC and the limit of detection referred to as the minimum detectable activity. All relevant uncertainties were propagated into the LC and minimum detectable activity. In addition, detection limits were calculated for the purpose of Safe Drinking Water Act monitoring. The detection capabilities were compared with the regulatory requirements in the US and EU for food and water. For samples spiked with either pure 89Sr or 90Sr, false positives were observed for the opposite radionuclide exceeding the above LC values. This was attributed to interference by the spiked activity. A new method was developed to calculate decision and detectability curves in the presence of interference.