The current study evaluated the psychometric properties of Wiseman and Watt's (2004) negative and positive superstitious belief items. The original items were compared with a modified, reworded set of items which emphasized each item's relation with either good or bad luck, and standard psychometric analyses were done. Modifying the items did not improve their psychometric properties; there was a negligible effect on Cronbach alpha, and Positive Item 3 continued to perform poorly. Confirmatory factor analysis, using the maximum likelihood method, suggested that a two-factor solution was preferable to a one-factor solution for both the original and modified items and that the problematic item should be discounted. It was concluded that the items require development and refinement before firm conclusions can be made about the factorial structure of superstitious belief. These results should also be tested further using Rasch methods.