Your privacy concern is not necessarily undue paranoia. It can be difficult to adequately anonymize genetic data, which can be personally identifying by its ver

Your privacy concern is not necessarily undue paranoia. It can be difficult to adequately anonymize genetic data, which can be personally identifying by its very nature, without much accompanying metadata. And you should also keep in mind that you can partially compromise the privacy of your genetic relatives as well. 23 and me may have reasonable privacy protections for the current state -of-the-art, but advances in genomics mean that what is 'anonymized' today might be personally identifiable in the near future. And what happens if 23 and Me, or its assets, are sold in the future? Unlike email addresses, login details, or bank accounts, genetic data remain personally identifying for you and your relatives and descendants and are inalienable (at least until the BioSingularity) While it is true that anyone could collect your shed cells to analyze, collecting personal artefacts requires some degree of prior specific interest in you sufficient to justify the costs of the collection and analysis - this is a barrier. When your data is uploaded into an easily-searchable database, this barrier is lowered, allowing those with access to trawl the DB for any members matching certain criteria. You should also keep in mind that all of these private databases are open to subpoena by various governmental jurisdictions. If you are familiar with the Birthday Problem and the various common abuses of Bayesian reasoning, you can imagine how this could pose a problem even for law-abiding citizens (and their relatives, remember). Additionally, it should be noted that regulations against adverse use of publicly-stored genetic information (e.g. by insurance companies) are by no means universal. Aggregated SNP data is very useful for researchers. But the utility of these commercial SNP products for the customer - beyond satisfying personal curiosity - is often overstated. If you are in the fortunate position of having access to a reasonably adequate family medical history and regular physical examinations, you will receive little marginal medical benefit from these services. Many of the phenotypes that customers spend the most time discussing on various fora can be reconstructed with reasonable accuracy from a pedigree, and the predictive power of these assays for many polygenic traits is fairly low at present. If you are interested in a few particular traits (like caffeine or drug metabolism) I would suggest ordering a kit for just those traits at first, and also playing around with data available on the 1000 Genomes website (www.1000genomes.org). This can be a bit more expensive, but it allows you to establish your comfort level with DNA data before exposing you to any privacy concerns. Of course, if you are interested in conducting personal research on your own SNP chip data and are already familiar with the data format and all of the associated statistical caveats, 23 and Me provides a good product at a reasonable price.