Previous efforts to investigate the equivalence of rating sources for job analysis ratings have reported conflicting results. In the present research, correlational and generalizability analyses were conducted to examine the equivalency of rating sources for over 70 state civil service job classifications. Incumbent and supervisor ratings (N=697) were examined, along with ratings of 3 trained experts derived from narrative job descriptions. Separate survey instruments were individually developed to obtain objective skills and abilities for each job classification. Pearson correlation coefficients reported significant reliability for ratings across each rating source, and significant convergent validities across sources. Results of the generalizability analyses were inconclusive concerning the similarity of ratings, however, variance attributed to rating sources was negligible. The results indicated that rating source did not have a significant effect on job analysis ratings when using an ability-oriented instrument. Ratings provided by incumbents, supervisors, and experts were similar in 65 of the 70 analyses. These findings support Smith and Hakel's (1979) belief that rating source does not make any practical difference in job analysis rating. (NB)