ABSTRACT A semi-analytic approximation is derived for the time-dependent fraction $f_\gamma (t)$ of the energy deposited by radioactive decay $\gamma$-rays in a homologously expanding plasma of general structure. An analytic approximation is given for spherically symmetric plasma distributions. Applied to Kilonovae (KNe) associated with neutron stars mergers and Type Ia supernovae, our semi-analytic and analytic approximations reproduce, with a few per cent and 10 per cent accuracy, respectively, the energy deposition rates, $\dot{Q}_{\rm dep}$, obtained in numeric Monte Carlo calculations. The time $t_\gamma$ beyond which $\gamma$-ray deposition is inefficient is determined by an effective frequency-independent $\gamma$-ray opacity $\kappa _{\gamma ,\text{eff}}$, $t_\gamma = \sqrt{\kappa _{\gamma ,\text{eff}}\langle \Sigma \rangle t^2}$, where $\langle \Sigma \rangle \propto t^{-2}$ is the average plasma column density. For $\beta$-decay dominated energy release, $\kappa _{\gamma ,\text{eff}}$ is typically close to the effective Compton scattering opacity, $\kappa _{\gamma ,\text{eff}} \approx 0.025$ cm$^{2}$ g$^{-1}$ with a weak dependence on composition. For KNe, $\kappa _{\gamma ,\text{eff}}$ depends mainly on the initial electron fraction $Y_e$, $\kappa _{\gamma ,\text{eff}} \approx 0.03(0.05)$ cm$^{2}$ g$^{-1}$ for $Y_e \gtrsim (\lesssim) 0.25$ (in contrast with earlier work that found $\kappa _{\gamma ,\text{eff}}$ larger by 1–2 orders of magnitude for low $Y_e$), and is insensitive to the (large) nuclear physics uncertainties. Determining $t_\gamma$ from observations will therefore measure the ejecta $\langle \Sigma \rangle t^2$, providing a stringent test of models. For $\langle \Sigma \rangle t^2=2\times 10^{11}~{\rm g\, {cm}^{-2}\, s^2}$, a typical value expected for KNe, $t_\gamma \approx 1$ d.