The Follow-up Study of a National Cohort of Gulf War and Gulf War Era Veterans is a multimode web, mail, and CATI survey. The original cohort for this longitudinal survey was comprised of 15,000 deployed Gulf War Veterans and 15,000 non-deployed Gulf War Era Veterans surveyed in 1995-1997 and then again in 2005. For the Wave 3 follow-up survey conducted in 2013, we used response propensity modeling (multiple logistic regression analysis) to examine the nonresponse (attrition) mechanism. Our assessment shows that the nonresponse weight adjustment via calibration successfully removed about 80% of the nonresponse bias in the test variable - marital status in 1991 as recorded in the frame data. It also significantly reduced the nonresponse bias initially observed in three of the five key Wave 3 survey variables examined: Unexplained Multisymptom Illness, smoking, and alcohol use. However, as evidenced by significant correlations between the examined outcomes and response propensities, we were unable to reduce the nonresponse bias for two of the five key variables: PTSD symptoms and general health. The negative correlation of PTSD symptoms with response propensity indicates that respondents tend to have lower PTSD screening scores (i.e., screen negative for PTSD) compared with nonrespondents. The positive correlation of response propensity with general health indicates that respondents tend to have higher self-reported general health compared with nonrespondents. This paper will describe the methodology used in the nonresponse bias analysis and discuss the performance of the nonresponse correction and its meaning for the results.