Objective: In this paper we present lessons learned from an online study assessing couples' health behaviors. Methods: We conducted an online cross-sectional study to assess health behaviors of dual-smoker couples. Participants were recruited via passive and targeted methods. Data were collected from 77 (pre-safeguard) and 197 (post-safeguard) participants. Safeguards included: (1) changing the incentive from prepaid card to raffle; (2) allowing only one IP address per response; (3) masking eligibility; (4) adding multiple questions to ensure consistency in responses; and (5) emphasizing data surveillance. We computed descriptive statistics using SAS 9.4 to compare enrollment rates and validity of data between the pre- and post-safeguard participants. Results: Although 77 entries were collected within 24 hours (presafeguards), 5 responses were ineligible and excluded. Among the remaining 72 entries, 68.1% were fraudulent as either multiple data entries (24.5%) and/or conflict in responses to similar survey items (83.7%). Once safeguards were administered (post-safeguards), data collection took longer to obtain 297 participants, which included 27 ineligibles. Among the 270 eligible participants, 35.9% were fraudulent due to conflicting responses to similar survey items. Conclusion: Online data collection via surveys should use safeguards to capture valid data. Many safeguards exist, which researchers should consider when designing online survey projects.