Watson and Rayner's 1920 experiment known as "Little Albert" is considered a classic in the history of psychology. It is frequently cited as the first and strongest demonstration of how fear can be learned by Pavlovian conditioning. Some counter-conditioning techniques derived from this experiment are used until now in phobias treatment. However, it is possible to identify differences in way this research is presented in literature (a reference of experimental rigour and creative interpretation) and the original article (with many methodological problems and theoretical inconsistencies). The objective of this paper is to present and to make a critical analysis of the cited article, identifying its qualities and limitations. Some problems of procedure and interpretation are discussed, for example: use of positive punishment, in addition to the programmed stimuli pairing, and the lack of consistent data about the acquisition of generalized fear. Methodological fails are also confirmed in unsuccessful replications of the experiment made during 1920s and 1930s. Moreover, this paper discusses the possible origin of the problems, the mistakes solidify by secondary literature and the real contributions of this study.