There’s very good agreement among 23andme and decodeme genetic profiles.
Antonio C B Oliveira had himself tested and wrote a little program to compare the results, finding only 23 discrepancies out of 560299 calls made by both services. Megan Smolenyak’s husband’s tests from both services were compared by Ann Turner, who found 35 disagreements among 560128 co-calls.
The called disagreements are fewer than the no call differences, which certainly seems like an acceptable rate to me. I wonder what the disagreements would be if someone were to have themselves tested twice, using two independent samples, say 6 months apart?
The Genetic Genealogist has a nice summary.