Amazon Mechanical Turk (MTurk) is a website used to crowdsource tasks, and for a decade it has been used to conduct research within the fields of social science. This paper examines the efficacy of using MTurk to recruit people with disabilities for survey samples, in contrast to the traditional method of working through established relationships with disability organizations. The study finds that each approach can over- and under-represent different groups within the general population and suggests that a combination of both methods may create the most diverse and accurate sample.