Various factors which influence the relationship between the Rasch item characteristic curve and the actual performance of an item are identified. The Rasch item characteristic curve is a new concept in test design and analysis. The Rasch test model provides information concerning the percent of students with a specified achievement level who would be expected to correctly answer a question with a specified difficulty. Using this information, it is possible to make a continuous plot of the expected percent correct across the full range of achievement. This plot constitutes the Rasch item characteristic curve. From the data gathered in this study it appears that using a minimum score group size of three to five students, a refined fit cutoff of .05 to .10 are optimum. Future research will be directed toward cross validating these findings. (Author/RC)