Dimensional preference for either form or color has been studied for a long time. Recently, many theor ists have assumed that dimensional preference is an attentiona lParameter with which learing is intimatelyassociated. The purporses of this study are as follows: 1. In experiment I, reliability of dimensional preference is measured through the “modified triads method” This method differs from the triads method (Brain and Goodenough, 1929) in the presentation of the stimulus. 2. In experiment II, psychological validity of estimation by optional shifts method is measured through the “modified optional shifts method”. This method differs from the optional shifts method (Smiley & Weir, 1966) in the trials by Which the subject's retention of discrimination learning is confirmed. That is, the, subject is given no reinforcement in the trials of the “modified optional shifts method ” The Ss, common to both experiments, are 21 Kindergarten children, 14 second-grade children, 18 fourth-grade children, and 10 undergraduates of both sexes,(TABLE1) Experiment I Measurement by the modified triads method is repeated twice with a one day period between trials. The results shown in TABLE2 indicates that (a) reliability is high in kindergarten children and undergraduates, and (b) most second-grade Ss match consistently, on form in a highly stable manner. Experiment II Validity of estimation in examined by the number of errors in the trials by. which Ss have been confirmed in keeping discrimination learning. The results (without undergraduates) are shown in TABLE 3, 4, 5, and 6, indicate that the unmber of times Ss match on form or color is a function of the number of errors in confirmation trials. Consequently, it is concluded that the number of times Ss match on form or color is not considered to be an index of dimensional preference. The results in undergraduates, compared with fourth-grade children, shown in TABLE 7, indicate that there is no differenece betwen undergraduates and fourth-grade children in the number of times Ss match on form or color. But further analysis of Ss, response patterns yields a developmental difference between these two grades.