When designing trials, an important ethical and practical concern is ensuring that proper consideration is given to patient preferences. A number of design variants on the traditional randomized trial can take account of such preferences in different ways. The partially randomized patient preference (PRPP) design is often advocated for taking these preferences into account. In studies of this type, all eligible patients are followed up, but the primary comparison is still between effects observed in patients randomly allocated to treatment arms. This design and other variants on conventional trials each have different scientific, ethical, and practical implications. We consider these diverse aspects of several modifications to the conventional method for obtaining consent and then, in more detail, the relative merits of the PRPP design. We conclude that none of these adjustments is ideal. Those designing trials should therefore always think carefully about the consent process, for its implications on participation and on how the results will be interpreted.