The authors evaluated principles of macroeconomics textbooks for readability using Coh-Metrix, a computational linguistics tool. Additionally, they conducted an experiment on Amazon's Mechanical Turk Web site in which participants ranked the readability of text samples. There was a wide range of scores on readability indexes both among textbooks and within textbooks. Results from the Mechanical Turk experiment revealed that the Flesch Reading Ease Index does not predict which samples readers will prefer, but readers do prefer samples that are thematically similar, as identified by Latent Semantic Analysis. There were differences in the responses of native and non-native-but-proficient English speakers to the text samples, suggesting that the intended audience is an important determinant of readability. (Contains 3 tables and 4 notes.)