BACKGROUND: The feasibility of randomized trials often depends on successful patient recruitment. Although numerous recruitment barriers have been identified it is unclear which of them complicate recruitment most. Also, most surveys have focused on the patients' perspective of recruitment barriers whereas the perspective of recruiting physicians has received less attention. Therefore, our aim was to conduct a postal survey among recruiting physicians of a multi-center trial to weigh barriers according to their impact on recruitment. METHODS: We identified any potential recruitment barriers from the literature and from our own experience with a multi-center trial of respiratory rehabilitation in patients with chronic obstructive pulmonary disease. We developed and pilot-tested a self-administered questionnaire where recruiting physicians were asked to express their agreement with statements about recruitment barriers on a Likert-type scale from 1 (full agreement with statement = very substantial recruitment barrier) to 7 (no agreement with statement = no recruitment barrier). RESULTS: 38 of 55 recruiting physicians returned questionnaires (69% response rate), of which 35 could be analyzed (64% useable response rate). Recruiting physicians reported that "time constraints" (median agreement of 3, interquartile range 2-5) had the most negative impact on recruitment followed by "difficulties including identified eligible patients" (median agreement of 5, IQR 3-6). Other barriers such as "trial design barriers", "lack of access to treatment", "individual barriers of recruiting physicians" or "insufficient training of recruiting physicians" were perceived to have little or no impact on patient recruitment. CONCLUSION: Physicians perceived time constraints as the most relevant recruitment barrier in a randomized trial. To overcome recruitment barriers interventions, that are affordable for both industry- and investigator-driven trials, need to be developed and tested in randomized trials. TRIAL REGISTRATION: ISRCTN84612310.