A study was designed to evaluate the effects of anticipating the matching response on acquisition and transfer of a second-order matching-to-sample task. 40 college students of both sexes were assigned to different anticipation procedures that involved several verbal descriptions of the correct response, as well as others not involving explicit descriptions. All subjects were exposed to a pretest, two training blocks, two transfer test blocks, and a posttest. Anticipation procedures were evaluated during the first training block using an observational training procedure, except for 8 subjects, who were exposed to two training control procedures. Analysis showed acquisition and transfer of successful matching seems to depend on a verbal discrimination of the matching criterion exemplified by second-order stimuli. This discrimination was facilitated by prior conditions, as when subjects showed correct matching performance at pretest or was established with anticipation procedures that required explicit reading of verbal descriptions of the response choice.