The popularity of online surveys is rising, yet the validity of survey data collected online is frequently questioned. This study compares online surveys versus paper surveys administered via regular mail in the tourism context, and examines in detail the extent and nature of survey bias resulting from survey format-specific respondent self-selection. Results suggest that (1) both online and mail samples deviate from census data population statistics regarding sociodemographics to the same extent (but differ in nature), (2) no differences exist in the contamination of data by response styles, (3) online respondents have a lower dropout rate and produce less incomplete data, and (4) responses to tourism-related questions differ significantly, indicating that survey format can dramatically influence results of empirical studies in tourism. Our findings show that neither pure online surveys nor pure paper surveys administered through regular mail are unbiased. Format-specific self-selection of respondents to participate leads to systematic biases in both cases, making multimethod survey approaches the most reliable way of data collection at present.