Is there a theory yet to be discovered that underlies quantum theory and explains its structure? If there is such a theory, one of the features it will have to explain is the central role of complex numbers as probability amplitudes. In this talk I explore the physical meaning of the statement “probability amplitudes are complex” by comparing ordinary complex-vector- space quantum theory with the real-vector-space theory having the same basic structure. Specifically, I discuss three questions that bring out qualitative differences between the two theories: (i) Is information about a preparation expressed optimally in the outcomes of a measurement? (ii) Are multipartite states locally accessible? (iii) Is entanglement “monogamous”?