LET  P1,…, PK BE INDEPENDENT UNIFORM POPULATIONS, U (0, QI) , I=1,…, K, RESPECTIVELY. SUPPOSE INDEPENDENT RANDOM SAMPLES OF EQUAL SIZE ARE DRAWN FROM EACH OF K POPULATIONS AND LET XI DENOTE THE LARGEST OBSERVATION OF THEI TH SAMPLE, I=1,…, K. THE POPULATION CORRESPONDING TO THE LARGEST XI IS SELECTED. WE CONSIDER THE PROBLEM OF ESTIMATING THE PARAMETERM OF THE SELECTED POPULATION UNDER THE SQUARED LOG Error LOSS FUNCTION. FORK ³2, WE OBTAIN THE UNIFORMLY MINIMUM RISK UNBIASED ESTIMATOR (UMRUE) OF QM. FOR K=2, SOME ADMISSIBILITY RESULTS FOR A SUBCLASS OF ESTIMATORS ARE DERIVED. WE SHOW THAT THE NATURAL ESTIMATOR X (2) IS INADMISSIBLE.