We consider discrete penalization schemes for reflecting stochastic differential equations. The convergence results obtained by Liu are generalized and refined. We also compare the penalization schemes with a more well-known recursive projection scheme.