Two tests used in the evaluation of the Conceptually Oriented Program in Elementary Science were analyzed by using the responses of 341 children. The COPES Test of Science Concepts and the COPES Test of Critical Terms were both shown to have a high content validity. Item analysis data showed that not all items in the tests discriminated satisfactorily at each grade level. Neither of the tests met the criteria for uni-dimensional scale when Green's Scalogram Analysis was used, and both were shown to be multi-dimensional by factor analysis. There was a significant correlation between both COPES tests and the criteria chosen. Recommendations for increasing the reliability of the tests, for increasing the original item pools, and for future test analyses were made. (Author/AL)