Abstract : This paper examines the asymptotic properties of compromise estimators. By this we mean an estimation method which compromises between a finite number of sampling situations in a small sample optimal way. We develop the asymptotic theory of such estimators and show that under a specific choice of sampling situations the compromise estimator is asymptotically robust in Huber's sense. Originator-supplied keywords include: Robust estimation, Conditional inference, Equivariance, Asymptotics.