What do clinical supervision research reviews across the last 25 years tell us? That question is subsequently examined. Based on database and literature searches, 20 reviews appearing from 1995 through 2019 were identified for survey examination; consistencies, inconsistencies and other defining features were determined across reviews; and the survey findings and their implications are considered. Primary findings are as follows: (a) ‘proof’ for supervision appears to be more ‘proof by association’ than otherwise, being primarily a product of ex post facto, cross‐sectional, correlational study; (b) evidence supporting supervision impact of any type is weak at best, especially so for worker and client outcomes; (c) supervision models generally lack empirical foundation; (d) evidence‐based supervision appears to be more a hope and dream than supervision‐based reality at present; and (e) the primary methodological problems that plagued supervision research in the 1990s are still all too frequent in modern research. Some questions to entertain about supervision going forward, and some remedies for improving its research, are proposed.