Propensity interpretations of objective probability have been plagued by many serious objections since their origin with Karl Popper (1957). In this talk, I attempt to offer a novel way of understanding propensity such that, in certain contexts, these problems go away. I call this approach, mechanistic propensity. The central claim is this: in cases where stochastic biological phenomena are the result of underlying mechanisms, propensities can be understood as properties of these mechanisms. I suggest that this (localized) account of propensity, if successful, enjoys several benefits that traditional accounts have lacked. Mechanistic propensities (1) aren't deeply mysterious (2) can explain frequencies, (3) are capable of accommodating single-case probabilities, (4) can cohere with determinism (both local and global), (5) avoid being too modal for science, and (6) seem not to face the reference class problem.