Gamma-ray spectrometry in the field was used in connection with an intensive study of the abundance and distribution of U and Th in peralkaline rocks of the Ilimaussaq intrusion, south Greenland. U and Th were determined from counts accumulated in the 1.76 and 2.62 MeV intervals of a portable gamma spectrometer.In order to permit a direct comparison with conventional surface sampling we define an effective sample in field gamma spectrometry as the volume of rock in which the variance of a particular radioelement (U, Th, or K) is equal to the estimation variance of a gamma -spectrometric determination of this radioelement. This volume contributes about 60% of the measured gamma rays. It is roughly one-tenth the volume which contributes 90%, the sample size which until now has been widely accepted. The mass of the effective sample for a Nal (T1) detector whose sides are surrounded by 5 cm of lead, placed on a rock surface, is approximately 20 kg and approximately 15 kg for 2.62 and 1.76 MeV gamma rays, respectively. Calibrations for U and Th were achieved by use of large concrete standards of known radio-element content, and having a gamma -ray mass absorption nearly equal to that of the rock to be measured. The calibration for U based on the concrete standards was more accurate than a calibration based on rock samples collected from measurement localities. The accuracy of the field gamma spectrometer, depending on standard errors associated with calibration equations, was about 2.5% for Th, and ranged from 2.5 to 7.5% for U, as Th/U varied from 1 to 5. Analytical precision for single measurements, combining the effects of counting statistics and instrument instability, ranged between 1 and 8% for Th, and between 1 and 12% for U. As many as 100 measurements per day with a lead-collimated detector were made along grids with 1-meter spacings in rock with a rough surface and an uneven distribution of radioelements. Results confirmed that specific rock types have characteristic U and Th contents. The measurement system was accurate enough so that where adequate radioelement contrasts existed, contours of U and Th corresponded closely to rock type and followed structural trends in the rock.