I defend the projection postulate against two of Margenau's criticisms. One involves two types of nonideal measurements, measurements that disturb and measurements that annihilate. Such measurements cannot be characterized using the original version of the projection postulate. This is one of the most interesting and powerful objections to the projection postulate since most realistic measurements are nonideal, in Margenau's sense. I show that a straightforward generalization of the projection postulate is capable of handling the more realistic kinds of measurements considered (...) by Margenau. His other objection involves the EPR (Einstein-Podolsky-Rosen) situation. He suggests that there is a significant potential for violations of the no-superluminalsignals requirement of the special theory of relativity, if projections occur in this situation and others like it. He also suggests that what is paradoxical about this situation disappears if the projection postulate is rejected. I show that it is not possible to use measurements on pairs of spatially-separated systems whose states are entangled to transmit information superluminally, and generalize this result to include nonideal measurements. I also show that EPR's dilemma does not really depend on the projection postulate. (shrink)