Abstract: Web-based clinical trials matching systems including breast cancer patients are expanding rapidly. However, limited data exist regarding the demographics and attitudes of breast cancer patients using the Internet to search for clinical trials information. Biases in patient populations could be introduced by recruiting patients to trials through the Internet. This study was designed to compare breast cancer patients in the radiation oncology clinic to those using the Internet to search for clinical trials information. A piloted questionnaire assessing demographics and attitudes regarding clinical trials was offered through the radiation oncology clinic at the University of Pennsylvania and on the OncoLink website (http://www.oncolink.org). The questionnaire consisted of 18 questions and was answered by a total of 157 patients with breast cancer. Breast cancer patients using the Web were more likely to be interested in clinical trials testing new drugs or therapies (71 [53%] versus 4 [17%], p = 0.002). More clinic patients indicated they would need a greater than 50% chance of benefiting from a trial (12 [52%] versus 33 [25%], p = 0.01) and a less than 10% potential for serious toxicity from a trial (15 [65%] versus 51 [38%], p = 0.02) for consideration of enrollment. African Americans were more likely than other races to have never used the Internet to search for cancer-related information (4 [40%] versus 18 [12%], p = 0.01), more likely to indicate that they need a greater than 50% chance of benefit to enroll in clinical trials (8 [80%] versus 37 [25%], p = 0.001), and less likely to be interested in clinical trials testing new drugs or therapies (1 [10%] versus 73 [50%], p = 0.01). Breast cancer patients have different attitudes regarding clinical trials based on race, Internet usage, and previous trial enrollment. Biases may be introduced with recruitment for clinical trials through the Internet. Radiation oncologists must consider these issues when offering clinical trials information through the Internet.