Three Canadian institutional animal care and use committees were observed over a 1-year period to investigate animal ethical evaluation. While each protocol was evaluated, the observer collected information about the final decision, the type of protocol, and the category of invasiveness. The observer also wrote down verbatim all verbal interventions, which were coded according to the following categories: scientific, technical, politics, human analog, reduction, refinement, and replacement. The data revealed that only 16% of the comments were devoted to the 3 (...) explicit ethical categories and that most of the comments were technical. However, the analysis revealed that ethical concerns were implicit in both scientific and technical comments; either ethical norms had been translated into scientific and technical language, or some of the scientific and technical comments had an impact on the ethical treatment of animals. The results are discussed in relation to previous nonobservational research that identified potential pitfalls and bias in animal ethical evaluation. (shrink)