We contribute to population ethics by proposing and axiomatizing rank-discounted critical-level generalized utilitarianism (RDCLU). Population ethics is needed for evaluation of policies, e.g., concerning climate change, where population size depends on the chosen policy. We show that critical-level generalized utilitarianism and (a version of) critical-level leximin are the limits of RDCLU for extreme values of the rank utility discount factor. Moreover, we establish how RDCLU avoids serious objections raised against other variable population criteria. In particular, it escapes both the Repugnant Conclusion and the Very Sadistic Conclusion (one of which critical-level generalized utilitarianism fails for any critical level).