A procedure was developed to detect differential item functioning (DIF) in a standardized essay test using the Partial Credit Model, the general polychotomous form of the Rasch model. Using a panel of experts in the writing process, hypothesized explanations for DIF at some score points were developed. Data for the study included averaged narrative and explanatory essay scores on the Maryland Writing Test for over 47,000 ninth-grade students from the 1989 annual administration. Student score records were randomly selected to create separate data files of about 1,200 White, Asian, Black, Hispanic, male, and female students each. DIF was investigated using a statistical procedure and via an expert panel review procedure. The statistical procedure involved calibrating prompts in separate subgroups and calculating Draba's DIF statistic. The panel review process and results appear to be reasonably helpful for prompt development, but less useful at the post-administration prompt analysis. Implications for research are discussed, and recommendations are made for applications of both procedures in other contexts. Five tables and two graphs summarize study findings. (SLD)