USING AN ANALYSIS OF JUROR SERVICE IN DANE COUNTY, WIS., FROM NOVEMBER 1977 THROUGH APRIL 1978, IT IS SHOWN THAT A JURY SELECTION PROCEDURE THAT USES REPLACEMENT (I.E., THE NAMES OF THOSE SELECTED FOR JURY PANEL ARE REPLACED IN THE 'WHEEL' AS SOON AS PRACTICABLE AND ARE AGAIN AVAILABLE WHEN ANOTHER RANDOM SELECTION IS MADE) PRODUCES A TYPICAL POISSON PATTERN OF MULTIPLE USE OF SOME PERSONS WHILE OTHERS ARE NEVER USED. THE USE OF THE REPLACEMENT SELECTION PROCEDURE IS BASED IN THE FALSE BELIEF THAT CONTINUED RANDOMIZATION DEVELOPS A BETTER MIX AND CROSS-SECTION THAN A SINGLE RANDOMIZATION. CONTINUED RANDOMIZATION, HOWEVER, HAS THE CONTRARY EFFECT BY NARROWING THE SELECTION TO THOSE WHOSE SELECTION IS GOVERNED BY THE MATHEMATICS OF CHANCE, WHICH NECESSARILY DEPRIVES MANY OF THE OPPORTUNITY TO SERVE AS JURORS. COURTS MADE AWARE OF THE ADVERSE EFFECTS OF OVER-RANDOMIZATION HAVE CORRECTED THE PROCESS BY ESTABLISHING A SYSTEM WITHOUT REPLACEMENT, IN WHICH EACH JUROR, SELECTED IN SOME RANDOM ORDER, GETS A FIRST CHANCE TO SERVE BEFORE ANYONE GETS A SECOND CHANCE. TABULAR AND GRAPHIC DATA ARE PROVIDED. (RCB)