The responses to the items composing psychometrics tests are often used to estimate the respondents’ latent trait levels. Although a larger number of items improves measurement validity, the effect of respondents’ fatigue on the response quality should be acknowledged for developing reliable measurement tools. This contribution presents an item response theory-based algorithm (denoted as Léon) able to shorten existing tests by concurrently accounting for the measurement precision of the abbreviated test and response fatigue of the respondents. A simulation study compares the performance of Léon in approximating the measurement precision that would be obtained from the full-length test in absence of response fatigue against that of another algorithm that does not account for the response fatigue during the selection process. Although, on average, the two algorithms select the same number of items, Léon provides a better approximation to the measurement precision of the full-length test than the other algorithm. Limitations of the study are discussed.