We calculate the lifetime of the deuteron with dimension-nine quark operators that violate baryon number by two units. We construct an effective field theory for |∆B| = 2 interactions that give rise to neutron-antineutron (n-n) oscillations and dinucleon decay within a consistent power counting. We calculate the ratio of the deuteron lifetime to the square of the n-n oscillation time up to next-to-leading order. Our result, which is analytical and has a quantified uncertainty, is smaller by a factor ≃ 2.5 than earlier estimates based on nuclear models, which impacts the indirect bound on the n-n oscillation time and future experiments. We discuss how combined measurements of n-n oscillations and deuteron decay can help to identify the sources of baryon-number violation.a Corresponding author, f.oosterhof@rug.nl At the classical level the standard model (SM) has two accidental global U(1) symmetries associated with baryon-number (B) and lepton-number (L) conservation [1][2][3]. At the quantum level only B − L is conserved, while B + L is anomalous. Since it can be expected that all global symmetries are only approximate, it is plausible that beyond-the-SM (BSM) physics violates B, L, and B − L separately. For instance, extending the SM with the only gauge-invariant dimension-five operator leads to violation of L by two units [1][2][3]. Additional B-and L-violating operators appear at the dimension-six level, while the first gauge-invariant operators that violate B by two units (|∆B| = 2) appear at dimension nine [4].The best limits on B-violating interactions come from the observed stability of the proton. The limit on its lifetime translates into a scale Λ |∆B|=1 > ∼ 10 13 TeV for grand unified theories [5]. Such energies are out of reach of colliders. However, models exist wherein B is only violated by two units and the proton is stable [6][7][8]. These interactions lead to the oscillation of neutral baryons into antibaryons, in analogy to strangeness-changing SM interactions that lead to kaon-antikaon oscillations. In particular, a neutron in a beam can oscillate into an antineutron [9] that annihilates with a nucleon in a target, producing several pions with a few hundred MeV of energy [10]. An ILL experiment sets a lower limit on the neutronantineutron (n-n) oscillation time of τ nn > 0.86 × 10 8 s ≃ 2.7 yr (90% C.L.) [11], which converts to a BSM scale Λ |∆B|=2 > ∼ 10 2 TeV, within reach of future colliders. An experiment at the European Spallation Source can improve τ nn by two orders of magnitude [12], probing regions of parameter space relevant for the observed baryon asymmetry of the Universe [13].Apart from "in-vacuum" n-n oscillations, |∆B| = 2 interactions also induce the decay of otherwise stable nuclei. A bound neutron can oscillate inside the nucleus into an antineutron, which then annihilates with another nucleon. Since a neutron and an antineutron have very different potential energies, the typical nuclear lifetime is far greater than τ nn [4]. Alternatively, two nucleons can annihilate directly....