Assessment of radiation exposure to drinking, surface, and groundwater and of the associated health risks calls for accurate and precise 226Ra and 222Rn measurements. One method that fits the bill is liquid scintillation counting (LSC), which allows measurements in one-phase (homogenous) or two-phase samples. The aim of our study was to compare the measurement efficiency with both variations in Niška Banja spa water, known for its elevated 222Rn content to get a better insight into the stability and behaviour of the samples and 226Ra interference in samples spiked with 226Ra with 222Rn measurement. 226Ra interference was more evident in homogenous, one-phase and much lower in two-phase samples. However, one-phase samples offer more accurate indirect 226Ra measurements. Water-immiscible cocktails (in two-phase samples) have shown a limited capacity for receiving 222Rn generated by Ra decay from the aqueous to organic phase when 222Rn/226Ra equilibrium is reached. We have also learned that samples with naturally high 222Rn content should not be spiked with 226Ra activities higher than the ones found in native samples and that calibration of two-phase samples can be rather challenging if measurements span over longer time. Further research would require much lower 226Ra activities for spiking to provide more practical answers to questions arising from the demonstrated phenomena.