The efficiency with which words can be identified is assessed in basic research by measuring identification reaction time or by determining the probability of a correct response when targets are presented tachistoscopically for very brief exposure durations. Subjects in this experiment‐‐eight high school students and one elementary student whose mean reading comprehension scores were at the 4.6 grade level‐‐received practice in making both speeded and tachistoscopic identifications of high and low frequency words. These two training conditions were compared to each other and to a third control condition where target stimuli were presented at a self‐paced rate with no experimental press for efficient processing. Comparisons were made on the time required to silently read simple sentences containing the trained words. Although reaction time training yielded faster identifications of high and low frequency words in isolation, differential transfer to sentence reading was not obtained. It is argued that the cost of using the experimental technology in classrooms remains to be justified.