To extend the NAA k 0 method to LEPD counting, procedures were elaborated for determining the full-energy peak detection efficiency ϵ p and the peak-to-total ratio P/ T (yielding the total efficiency ϵ t = ϵ p /( P/ T ) . ϵ p is needed for concentration calculation. ϵ p and ϵ t are needed to correct for true coincidences (including KX and LX lines) in case of small source-detector separation. First the “reference” efficiency curve (log ϵ p,ref vs log E) was determined with calibrated point sources (from 4.65 keV ( 139Ce) to 383.9 keV ( 133Ba)) measured at ∼ 16 cm distance from the LEPD. The accuracy is ∼ 1–2% for E > 11 keV, and ∼ 5% for E < 11 keV. In view of actual analyses, ϵ p,ref was converted to ϵ p,geo (i.e. at a small distance from the LEPD) based on the calculation of the effective solid angles Ω with the program SOLANG. The accuracy of this, checked with practically coincidence-free lines of point sources measured close to the LEPD, was ∼ 5% on the average. After identifying the origin of all the pulses in the spectra, a P/ T curve ( log P/ T vs log E) was constructed from results obtained for the most prominent lines of some nine sources (from 5.9 keV ( 55Fe) to 320.1 keV ( 51Cr)). P/ T was found to be practically independent of the source-detector separation. Its accuracy of ∼ 5% is satisfactory in view of the large error reduction. The results and conclusions reached here were obtained from experiments with two different LEPDs at the INW, Gent and the LNETI, Sacavém.