Clinical trials are often hindered by insufficient participation. Difficulties may arise if a clinician is generally reluctant to enroll patients in clinical trials, if the clinician is unwilling to enroll a patient in a particular trial, or if the patient refuses to participate. In our three-stage survey of clinicians treating breast cancer, we found that a favorable orientation to trials in principle can be attenuated when clinicians do not like the treatment regimens in a particular trial. We also found a lack of consensus as to appropriate treatment for breast cancer coupled with a high confidence by individual clinicians in their own treatment decisions (a phenomenon we term "micro-certainty/macro-uncertainty"). Accordingly, in the most controversial situations, it may be hardest to mount trials. Trial design is also complicated by the variation among clinicians in the importance they assign to various patient characteristics in making their treatment decisions. This variation can lead to difficulties in establishing patient subgroups that will be accepted by the clinical community. The trial designer will need to be alert to these considerations; efforts to build consensus on which data are necessary and which therapies are acceptable should improve the design and application of clinical trials.