We report on a study of potential sources of therapeutic misconception in early phase gene transfer research, examining how investigators and their consent forms represent the prospect for direct benefit. Our analysis demonstrates that even though half of PIs said they expected direct medical benefit for their subjects, they did not necessarily convey this to their subjects. What they reported telling subjects resembled what was written in their consent form, which suggests that, far from being irrelevant, the consent form is an influential component of the consent process. We also demonstrate that the language used to describe direct benefit in consent forms and PIs' discussions was mostly vague, ambiguous, and indeterminate about benefit, rather than clearly negative. This was especially true for cancer and vascular disease trials. Our respondents found the problem of balancing hopes and expectations, for themselves and for their subjects, extraordinarily challenging. In the current era, investigators face such challenges without consistent normative guidance or agreed-upon standards for how to talk about scientific promise and uncertainty in early phase trials. This dilemma cannot be effectively addressed by individual investigators alone, but must be acknowledged and openly discussed by the scientific community at large.