The Science Teaching Efficacy Belief Instrument-Preservice (STEBI-B) has been used in many studies to measure science teaching self-efficacy and outcome expectancy in preservice elementary teachers. Since its development in 1990, there have been no studies that have re-examined its internal validity and reliability. The purpose of this study was to do so. Two hundred ninety preservice elementary teachers participated in this study. The STEBI-B was administered at the beginning of science methods courses. A factor analysis established that the two subscales, Personal Science Teaching Efficacy Belief (PSTE) and Science Teaching Outcome Expectancy (STOE), on the STEBI-B were homogeneous, and loadings were comparable to those reported by Enochs and Riggs (1990). Two items on the STOE were found to exhibit cross-loading on the factor analysis, as well as low item-total correlations. These two items were modified and the revised instrument administered to 86 new participants. Both revised items loaded more clearly on the STOE subscale, and item-total correlations were stronger. Comparison of means analyses showed that gender, number of science courses taken, and school science experiences had significant associations with PSTE. The STEBI-B continues to be employed in hundreds of studies to measure science teaching self-efficacy. It is important to continue to monitor its reliability and validity, as well as check on associations with various background variables. The modified STEBI-B is included in an appendix with this article.