A year-long study examined the effects of a computer-based reading project on students' attitudes and performance in reading and writing. Subjects (73 "at-risk" eighth-grade students in an inner-city school in Paterson, New Jersey) were randomly divided into control and experimental groups. The experimental group used a software package that dealt with the real-life issues of this special population. The control group used a variety of skill-based reading packages from the Minnesota Educational Computing Consortium as well as test preparation software for the New Jersey statewide assessment test. Three pre-post assessment instruments were used: (1) a 22-item teacher-made attitudinal survey; (2) alternate forms of the Gates-MacGinitie Reading Tests; and (3) part 1 of the writing section of New Jersey's High School Proficiency Test. Analysis of variance procedures indicated that while there were no significant differences attributable to the group for reading and writing, the attitudes of the experimental group toward their work with computers and themselves as readers and writers improved significantly. (Seven tables of data are included.) (Author/RS)