A primary goal of this study was to determine the extent to which item difficulty was related to item position and, if a significant relationship was found, to suggest adjustments to predicted item difficulty that reflect differences in item position. Item response data from the Medical College Admission Test (MCAT) were analyzed. A data set was created containing statistical data from two uses of each of 9,099 items. Three models were investigated for describing the relationship between item position and the decrease in the adjusted percent passing: (1) a linear model; (2) a quadratic model; and (3) an exponential model. Results show that item statistics can be significantly related to item position even for tests that are not apparently speeded. Taken as a whole, the results suggest a more comprehensive model of item difficulty with the time spent on each item as a mediating variable. Further development of such a model should be based on data from computer-administered tests where response latency data can be collected and analyzed. Other areas for future research include empirical investigations of the relationship of item position effects to examinee performance level and to characteristics of preceding items in the form. (LMO)