Low-radioactivity argon sources are desired by the WIMP dark matter experimental particle physics community. Accurate understanding of the subsurface production rate of the radionuclide 39Ar is also necessary for argon dating techniques and noble gas geochemistry of the shallow and the deep Earth. Our new calculations of subsurface production of neutrons, 21Ne, and 39Ar (Šrámek et al., 2017) take advantage of the state-of-the-art reliable tools of nuclear physics to obtain reaction cross sections and spectra (TALYS) and to evaluate neutron propagation in rock (MCNP6). We discuss our method and results in relation to previous studies and show the relative importance of various neutron, 21Ne, and 39Ar nucleogenic production channels. Uncertainty in nuclear reaction cross sections, which is the major contributor to overall calculation uncertainty, is estimated from variability in existing experimental and library data. Depending on selected rock composition, on the order of 107–1010 alpha particles are produced in one kilogram of rock per year (order of 1–103 kg−1 s−1); the number of produced neutrons is 6 orders of magnitude lower, 21Ne production rate drops by an additional factor of 15–20, and another one order of magnitude or more is dropped in production of 39Ar. Calculated 39Ar production rates span a great range from 29 ± 9 atoms per kg-rock per year in the K–Th–U-enriched Upper Continental Crust to (2.6 ± 0.8) × 10−4 atoms per kg-rock per year in the Depleted Upper Mantle. Nucleogenic 39Ar production exceeds the cosmogenic production below ∼ 700 m depth in the Earth. Recent report by the DarkSide-50 Collaboration (Agnes et al., 2016) puts the concentration of 39Ar in Doe Canyon, SW Colorado, deep CO2 well gas at 1400±200 times lower compared to the atmospheric value. While it was argued that the Doe Canyon gas is derived from the Earth’s upper mantle, it shows some counter intuitive isotopic characteristics, such as extremely low 3He/4He suggesting crustal origin, while at the same time extremely low 39Ar/40Ar indicating a source low in K, Th, U abundances. As a possible solution to this puzzle, we envisage a sizeable (i.e., low surface to volume ratio) gas reservoir in the shallow crust where mantle gas, contaminated by crust-derived gases (4He, 21Ne, and N2), accumu- lates for sufficient time (> 104 years) so that the 39Ar activity drops to the observed low value. Overall, the noble gas observations (esp. of helium, neon, argon), both in terms of out- gassing rates from the Earth and the gas’ isotopic composition, present a major challenge to geoscientists who strive to formulate a coherent story of the Earth’s formation, evolution, and current state. A link between underground noble gas production and decay of long lived radionuclides (40K, 232Th, 238U) ties noble gas geochemistry to dynamic models of thermal evolution of the Earth and to questions about deep Earth’s composition and architecture. References: Agnes et al. (2016): “Results from the first use of low radioactivity argon in a dark matter search.” Phys. Rev. D, 93(8): 081101, doi:10.1103/PhysRevD.93.081101 (arXiv:1510.00702) Gilfillan et al. (2008): “The noble gas geochemistry of natural CO2 gas reservoirs from the Colorado Plateau and Rocky Mountain provinces, USA,” Geochim. Cosmochim. Acta, 72 (4), 1174–1198, doi:10.1016/j.gca.2007.10.009 Šrámek et al. (2017): “Subterranean production of neutrons, 39Ar and 21Ne: Rates and uncertainties.” Geochim. Cosmochim. Acta 196, 370–387, doi:10.1016/j.gca.2016.09.040 (arXiv:1509.07436) This presentation was used for the Low-Radioactivity Underground Argon Workshop held at Pacific Northwest National Laboratory in Richland, Washington on March 19 - 20, 2018.