Abstract "Semantic priming" has two meanings, one an internal activation and the other a performance difference between responses to primed and unprimed stimuli. Individual and group differences in the performance difference score are heavily influenced by overall speed and/or accuracy and so are a flawed reflection of internal activation. These relationships obscure the meaning of many published findings of greater than normal priming difference scores in special groups, such as children and the elderly. Meaningful comparison of groups on the activation underlying priming difference scores requires removing the effects of overall performance level. We discuss ways of doing so. We make this argument in the context of semantic priming research but it applies equally to other paradigms that use difference scores for the comparison of groups of differing performance levels.