RECENT RESEARCH ON THE DISCRIMINATION PROCESS INDICATES THAT ERRORLESS LEARNING CAN OCCUR IF STIMULI ARE CAREFULLY PROGRAMED SO THAT THEY ARE DISSIMILAR AND GRADUALLY BECOME MORE SIMILAR AS TRAINING PROCEEDS. TO ASSESS THIS APPROACH IN TEACHING THE LOWER- CASE LETTER B AND D WITH KINDERGARTEN SUBJECTS, TWO SETS OF STIMILI CONSTITUTED THE FIRST EXPERIMENTAL VARIABLE. IN THE PROGRESSIVE VALUE OF THIS VARIABLE, COLOR, SIZE, PRESENCE OF PROMPTS, AND DURATION OF PRESENTATION WERE PROGRESSIVELY FADED TO THE TERMINAL DISCRIMINATION. THE SECOND VALUE OF THIS VARIABLE WAS THE TERMINAL DISCRIMINATION IN THE PROGRESSIVE STIMULI--THE B AND D CONSTANT IN THE ATTRIBUTES OF COLOR, SIZE, DURATION OF PRESENTATION, AND ABSENCE OF PROMPTS. THE SECOND VARIABLE WAS THE TIME OF INTRODUCTION OF THE SECOND LETTER. TWO SEPARATE ANALYSES OF COVARIANCE REVEALED SIGNIFICANT EFFECTS FOR THE TIME VARIABLE, BUT NOT FOR THE PROGRESSIVE-CONSTANT VARIABLE. THE EARLY-PROGRESSIVE COMBINATION RESULTED IN 81 PERCENT OF THE SUBJECTS LEARNING THE DISCRIMINATION WITH UNDER 10 PERCENT ERRORS. WHEN SUBJECTS WERE CATEGORIZED INTO SAME, MIXED, OR CROSSED LATERAL DOMINANCE, NO DIFFERENCES IN ERRORS ON THE TASK WERE OBSERVED. THE SUBJECTS WHO LEARNED THE DISCRIMINATION WITHOUT ERRORS SUBSEQUENTLY COULD NOT DRAW THE LETTERS. FIFTY-FIVE REFERENCES ARE INCLUDED. (AUTHOR/KJ)