The present investigation employs item response theory (IRT) to develop an abbreviated Medical Fear Survey (MFS). Application of IRT analyses in Study 1 (n = 931) to the original 50-item MFS resulted in a 25-item shortened version. Examination of the location parameters also resulted in a reduction of the Likert-type scaling of the MFS by removing the last response category (“Terror”). The five subscales of the original MFS were highly correlated with those of the MFS–short version. The short version also displayed comparable convergent and discriminant validity with the original MFS in relation to measures of fear, disgust, and anxiety. Confirmatory factor analysis in Study 2 revealed that the five-factor structure of the MFS–short form fit the data well in U.S. (n = 283) and Dutch (n = 258) samples. The short form also had comparable convergent and discriminant validity with the original MFS in relation to domains of disgust in both samples. Receiver operating characteristic (ROC) analysis in Study 3 demonstrated that the subscales of the short version were comparable with the original MFS in classifying participants high (n = 40) and low (n = 40) in blood/injection phobia. Last, structural equation modeling in Study 4 (n = 113) revealed that the MFS–short form demonstrated excellent convergent/discriminant validity with strong associations with injection fear and no association with spider fear. These findings suggest that the MFS–short form has considerable strengths, including decreased assessment time, while retaining sound psychometric properties.