Although the evidence for the use of situational judgment tests (SJTs) in high-stakes testing has been generally promising, questions have been raised regarding the potential coachability of SJTs. This study reports the first examination of the effects of coaching on SJT scores in an operational high-stakes setting. We contrast findings from a simple comparison of SJT scores for coached and uncoached participants (posttest only) with three different approaches to deal with the effects of self-selection into coaching programs, namely using a pretest as a covariate and using two different forms of propensity score-based matching using a wide range of variables as covariates. Coaching effects were estimated at about 0.5 SDs. The implications for the use of SJTs in high-stakes settings and for coaching research in general are discussed.