Population-based research is enhanced by biological measures, but biological sampling raises complex ethical issues. The third British National Survey of Sexual Attitudes and Lifestyles (Natsal-3) will estimate the population prevalence of five sexually transmitted infections (STIs) (Chlamydia trachomatis, Neisseria gonorrhoeae, human papillomavirus (HPV), HIV and Mycoplasma genitalium) in a probability sample aged 16-44 years. The present work describes the development of an ethical approach to urine testing for STIs, including the process of reaching consensus on whether to return results. The following issues were considered: (1) testing for some STIs that are treatable and for which appropriate settings to obtain free testing and advice are widely available (Natsal-3 provides all respondents with STI and healthcare access information), (2) limits on test accuracy and timeliness imposed by survey conditions and sample type, (3) testing for some STIs with unknown clinical and public health implications, (4) how a uniform approach is easier to explain and understand, (5) practical difficulties in returning results and cost efficiency, such as enabling wider STI testing by not returning results. The agreed approach, to perform voluntary anonymous testing with specific consent for five STIs without returning results, was approved by stakeholders and a research ethics committee. Overall, this was acceptable to respondents in developmental piloting; 61% (68 of 111) of respondents agreed to provide a sample. The experiences reported here may inform the ethical decision making of researchers, research ethics committees and funders considering population-based biological sampling.