European Union Council Directive 2013/51/EURATOM recently sets out so-called indicator parameters for: radon, tritium and indicative dose of water intended for human consumption. The aim of this research was to elaborate an effective procedure for determination of radon and radium 226,228Ra isotopes (which are potentially the main contributors to the internal dose from drinking and cooking water) and to find the possible relationships between these radionuclides in underground water reservoirs and 222Rn concentration in the soil gas in their vicinity. The research was performed by applying a non-volatile and water-immiscible scintillation cocktail based on a pure diisopropylnaphthalene (Ultima Gold F: UGF), which allow for efficient radon extraction from 0.5 dm3 of water samples to 20 cm3 of scintillation phase and its direct determination with a detection limit of 5 × 10–3 Bq dm−3. The further preliminary concentration of 3 dm3 of crude water samples by evaporation to 0.5 dm3 samples led to the removal of all unsupported 222Rn activity and allowed the 226Ra determination via equivalent 222Rn detection after one-month samples storage using a low-background Triathler liquid scintillation counter in the α/β separation counting mode. Together with determination of 226Ra isotope in water samples, the simultaneous measurements of 228Ra and 222Rn radionuclides concentrations in water as well as 222Rn activity in the soil gas around the water supply sites were performed. The achieved limit of 226Ra detection was at a very low level of 10–3 Bq dm−3. The measured values of 226Ra concentration in 50 public underground water supply units for the Kalisz district of Poland were relatively low and ranged from below detection limit to 28.5 × 10–3 Bq dm−3 with arithmetic mean and median values of 12.9 and 12.2 × 10–3 Bq dm−3, respectively. Weak correlations were observed between activity concentrations of 226Ra and 222Rn in the crude water samples (R2 = 0.31) and 222Rn in water and its concentration in the nearby soil gas (R2 = 0.48).