Field trials conducted in multiple years across several locations play an essential role in plant breeding and variety testing. Usually, the analysis of the series of field trials is performed using a two-stage approach, where each combination of year and site is treated as environment. In variety testing based on the results from the analysis, the best varieties are recommended for cultivation. Under a Bayesian approach, the variety recommendation process can be treated as a formal decision theoretic problem. In the present study, we describe Bayesian counterparts of two stability measures and compare the varieties in terms of the posterior expected utility. Using the described methodology, we identify the most stable and highest tuber yielding varieties in the Polish potato series of field trials conducted from 2016 to 2018. It is shown that variety Arielle was the highest yielding, the third most stable variety and was the second best variety in terms of the posterior expected utility. In the present work, application of the Bayesian approach allowed us to incorporate the prior knowledge about the tested varieties and offered a possibility of treating the variety recommendation process as a formal decision process.