A test of job-specific literacy, called the McCoy Assessment of Worksite Literacy Abilities (MAWSLA), could not predict the job success of the 100 employees who took it. The 108-item test was based on literary tasks encountered by employees in the course of a routine day at McCoy Electronics Company, a 560-employee firm in Pennsylvania. The instrument was developed after observation; collection of literary samples that employees routinely had to read (e.g., memos from management, safety instructions, personnel forms, and operating instructions); interviews with workers, supervisors, and management; and two pilot tests. After taking the 30-minute MAWSLA, subjects took the Nelson Denny reading tests. Multiple linear regression determined the extent to which MAWSLA scores accounted for variance in job performance ratings and in an "employee value index," constructed from performance ratings, time missed from work, and number of days on the job. Other variables added to the prediction formula were age, length of employment, and scores on the Nelson Denny tests. The analysis accounted for only 6 percent of variance in the employee value index and only 10 percent in job ratings. Use of the Pearson-r revealed no correlation between the MAWSLA and the index, and a correlation of only .21 between the MAWSLA and performance ratings. (CML)