Subjects carried out a choice reaction time experiment in which they were required to press one lever if a tachistoscopically flashed numeral were a member of a previously memorized list of numerals and a second lever if it were not a member of the list. The number of numerals in the list was varied from one to four. On half of the trials the numeral was partially masked by a pattern of visual noise. For one group a single fixed pattern of noise was used; for a second group four patterns presented in random order were used. Reaction time from the onset of the numeral to the lever press was found to be a linear function of list size, confirming Sternberg's hypothesis that the list is scanned serially to determine a match. Partial masking was found to add a constant increment to reaction time for both groups, independent of set size. The increment was equal for both groups. The constancy of the reaction time increase contradicts Sternberg's results in which the increment was found to be, in part, a function of list size. The results of the present study indicate that one cannot dismiss the possibility that subjects identify the character to the point of labelling prior to an attempt to determine list membership.