We examined cognitive predictors of speech and articulation rate in 50 individuals with multiple sclerosis (MS) and 23 healthy controls. We measured speech and articulation rate from audio-recordings of participants reading aloud and talking extemporaneously on a topic of their choice (i.e., self-generated speech). Articulation rate was calculated for each speech sample by removing lexically irrelevant vocalizations and pauses of >200 ms. Speech rate was similarly calculated including pauses. Concurrently, the Minimal Assessment of Cognitive Function in Multiple Sclerosis (MACFIMS) battery, as well as standardized tests of sentence intelligibility and syllable repetition were administered. Analysis of variance showed that MS patients were slower on three of the four rate measures. Greater variance in rate measures was accounted for by cognitive variables for the MS group than controls. An information processing speed composite, as measured by the Symbol Digit Modalities Test (SDMT) and the Paced Auditory Serial Addition Test (PASAT), was the strongest predictor among cognitive tests. A composite of memory tests related to self-generated speech, above and beyond information processing speed, but not to oral reading. Self-generated speech, in this study, was not found to relate more strongly to cognitive tests than simple reading. Implications for further research are discussed. (JINS, 2012, 18, 1–8)