ABSTRACT: Background:Individuals with primary progressive aphasia (PPA) and their caregivers want to know what to expect so that they can plan support appropriately. The ability to predict decline in naming and semantic knowledge, and advise individuals with PPA and their caregivers regarding future planning, would be invaluable clinically. Aims:The aims of this study were to investigate patterns of decline in naming and semantic knowledge in each of the clinical variants of PPA (logopenic variant PPA, lvPPA; nonfluent agrammatic PPA, nfaPPA; and semantic variant PPA, svPPA) and to examine the effects of other variables on rate of decline. We hypothesized that speech-language rehabilitation, higher education, and higher baseline test scores would be associated with slower decline, and older age with faster decline. Methods and Procedures:A total of ninety-four participants with PPA underwent language testing, including thirty six participants with lvPPA, thirty-one participants with nfaPPA, and twenty-seven participants with svPPA. All participant groups were similar in age and education. We focused on decline on three tests: the short form of the Boston Naming Test (BNT), the Hopkins Assessment of Naming Actions (HANA), and the short form of the Pyramids and Palm Trees Test (PPTT). Outcome and Results:Across language tests, the most precipitous rates of decline (loss of points per month) occurred in nfaPPA, followed by svPPA, then lvPPA. Female sex, longer symptom duration, higher baseline test score, and speech-language rehabilitation were associated with slower decline. Conclusions:PPA variants were distinguishable by rapidity of decline, with nfaPPA having the most precipitous decline. As hypothesized, higher baseline test scores and speech-language rehabilitation were associated with slower decline. Surprisingly, age and education were not important prognostically for individuals in this study. Further study of prognostically-relevant variables in PPA is indicated in this population.