While survey studies have examined bias much is unknown regarding specific subpopulations, especially women workers. METHODS: A population based phone, Internet, and mail survey of workplace falls during pregnancy was undertaken. Participation by industry and occupation and survey approach and bias, reliability, and incomplete data were examined. RESULTS: Of the 3,997 women surveyed, 71% were employed during their pregnancy. Internet responders were most likely to be employed while pregnant and to report a workplace fall at 8.8% compared to 5.8% and 6.1% for mail and phone respondents. Internet responders had the most missing employment data with company name missing for 17.9% compared to 1.3% for phone responders. Mail surveys were best for recruiting those employed in eight of nine industries, and this was especially true for service occupations. CONCLUSIONS: To decrease bias and increase participation, mixed approaches may be useful with particular attention for collecting occupational data.