Do good readers embody more or less? The current investigation examined embodiment effects as a function of individual reading skill in the context of two cognitive theories of reading comprehension. The Construction-Integration model predicts that sensorimotor activity during reading will correlate negatively with reading skill, because good readers focus on relations among abstract ideas derived from the text. Supposedly those abstract ideas have little or no sensorimotor content, hence any sensorimotor activity while reading is wasted effort. In contrast, the simulation theory predicts that sensorimotor activity will correlate positively with reading skill, because good readers create a simulation of what is happening within the text to comprehend it. The simulation is based in neural and bodily systems of action, perception, and emotion. These opposing predictions were tested using the reading-by-rotation paradigm to measure embodiment effects. Those effects were then correlated with reading skill measured using the Gates-McGinite standardized reading test. Analyses revealed an unexpected interaction between condition and congruency, and a negative relationship between embodiment and reading skill. Several caveats to the results are discussed.