The activity of environmental samples is usually measured by high resolution HPGe gamma spectrometers. In this work a set-up with a 9in.x9in. NaI well-detector with 3in. thickness and a 3in.×3in. plug detector in a 15-cm-thick lead shielding is considered as an alternative (Hansman, 2014). In spite of its much poorer resolution, it requires shorter measurement times and may possibly give better detection limits. In order to determine the U-238, Th-232, and K-40 content in the samples by this NaI(Tl) detector, the corresponding photopeak efficiencies must be known. These efficiencies can be found for certain source matrix and geometry by Geant4 simulation. We found discrepancy between simulated and experimental efficiencies of 5-50%, which can be mainly due to effects of light collection within the detector volume, an effect which was not taken into account by simulations. The influence of random coincidence summing on detection efficiency for radionuclide activities in the range 130-4000Bq, was negligible. This paper describes also, how the efficiency in the detector depends on the position of the radioactive point source. To avoid large dead time, relatively weak Mn-54, Co-60 and Na-22 point sources of a few kBq were used. Results for single gamma lines and also for coincidence summing gamma lines are presented.