[en] A series of benchmark tests has been performed since 1980 in order to examine the capability of JAERI's criticality-safety evaluation code-system JACS. This report presents the results of benchmark calculations on fissile array experiments with the Monte-Carlo code KENO-IV and the multigroup data library MGCL as a part of the JACS code-system. The experimental systems selected from USA publication vary in moderating ratio, physical/chemical property of nuclear fuel, reflecting condition, shape/size of array unit, distance between units, array size, from case to case. The calculated k_eff's for 108 cases in all distribute from 0.945 to 1.025. This distribution width is presumably due to the bias-errors in the calculated k_eff's, since the KENO-IV statistical error, some 0.007 at maximum standard deviation, predicts 1/2 or 1/3 of the width. The bias-errors in calculated k_eff's are acknowledged to be caused by the variation of the moderating ratio, the reflecting condition together with the distance between units. (author)