Protective measurement is a new measuring method introduced by Aharonov, Vaidman, and Anandan, with the aim of measuring the expectation value of an observable on a single quantum system, even if the system is initially not in an eigenstate of the measured observable. According to these authors, this feature of protective measurements favors a realistic interpretation of the wave function. These claims were challenged by Uffink. He argued that only observables that commute with the system's Hamiltonian can be protectively measured, (...) and that an allegedly protective measurement of an observable that does not commute with the system's Hamiltonian does not actually measure this observable, but rather another related one that commutes with the system's Hamiltonian. In this paper we identify a number of unresolved issues in Uffink's proofs and argue that his alternative interpretation of what happens in a protective measurement has not been justified. (shrink)