OBJECTIVES: The NIH roadmap project “Patient-Reported Outcomes Measurement Information System” (PROMIS) provides a basis for measuring symptoms - including fatigue; associated with chronic diseases.  However, information on use of PROMIS with Fibromyalgia (FM) patients is limited.  This study sought to re-estimate the Item Response Theory (IRT) item parameter estimates for each of the 95-items included in the PROMIS Fatigue item bank using responses from individuals with FM.  It is unknown whether the parameter estimates differ among a sample of FM patients from that of the broader population used in the initial PROMIS item calibrations. METHODS: An internet-based survey was conducted with individuals with FM.  Participants had to be ≥18 years and diagnosed with FM by a physician.  Respondents were randomized into one of three cohorts, each of which completed one-third of the 95-items from the entire PROMIS v1.0 Fatigue Item Bank (www.nihpromis.org).  IRT model fitting of Samejima’s Graded Response Model (GRM) was conducted using MULTILOG7 to determine item parameter estimates and compared with standard PROMIS parameter estimates. RESULTS: A total of 1,207 respondents completed the survey.  IRT item parameter estimates (slope and item characteristic curve thresholds) and item information functions were obtained for all 95-items.  Mean score across all items for the FM sample was 3.68 compared to the PROMIS mean of 2.12.  Maximum information function was >2 for 22/95 items in the FM sample, compared to only 2/95 items in the PROMIS sample, indicating differences in both discriminate ability as well as the difficulty of items for a FM specific population.  CONCLUSIONS: Disease specific item level calibrations of PROMIS items can be estimated.  Using the same GRM methodology as PROMIS we were able to calculate revised item parameter estimates and information functions.  These estimates can be used to construct improved efficient short-forms or for application in computer adaptive testing.