The application of statistical algorithms in the problem of early detection of technogenic, nuclear power plant emissions in the atmosphere near the ground by means of the spectrometric channel of ARMS (Automatic Radiation Monitoring System) is examined. It is noted that the changes in the radon progeny content which are associated with the diurnal behavior of the intensity of turbulent mixing influence the variance of the readings in the 131 I and 137 Cs detection ranges. As a result, when a stationary Poisson model is used in the algorithm the number of false alarms increases and the algorithm becomes ineffective. Regional corrections lowering the probability of false alarms to ≤10 -3 with correct detection probability ≥0.7 were introduced into the detection and classification algorithm on the basis of the results of monitoring performed over the course of a year for intervals with stationary γ background.When the ARMS (Automatic Radiation Monitoring System) spectrometric channel is used for early detection of nuclear power plant emissions, technogenic radionuclides at low concentrations must be detected with a definite reliability in real-time. Ordinarily, the lowest measurable activity, determined by the background conditions in the monitoring zone and the method and means for performing the measurements, is taken as the conventional boundary between the detection and measurement processes. Our assessments have shown that a gamma spectrometer with a 63 × 63 mm NaI(Tl) crystal with exposure time 15 min can be used for quantitative measurements of 131 I and 137 Cs provided that their volume activity ≥8 Bq/m 3 , i.e., starting at values 2-3 times lower than the admissible volume activity AVA PUBLIC according to . Statistical algorithms for processing the spectrometric information make it possible to detect 131 I and 137 Cs with probability ≥0.7 at concentrations concentration 2-3 Bq/m 3 , which expands the possibility of using the ARMS spectrometric channel for monitoring nuclear power plant emissions. In what follows, the term "detection" is used precisely in this probabilistic sense, since, for example, according to Rosgidromet normative documents the detection of a radionuclide not present in the global radiation background is taken as a level of detection two times above the minimum measurable activity of the radionuclide for the apparatus being using [2].The concept of the minimum detectable volume activity can be introduced as a characteristic of the spectrometric channel for detecting technogenic radionuclides at low concentrations under background conditions in the ARMS monitoring zone. It can be determined as the volume activity which is detected by the algorithm used for processing the spectrometric information with probability of correct detection P c.d and false-alarm probability P f.a established for the given ARMS.