Behavioral tests have become important tools for the analysis of functional effects of induced mutations in transgenic mice. However, depending on the type of mutation and several experimental parameters, false positive or negative findings may be obtained. Given the fact that molecular neurobiologists now make increasing use of behavioral paradigms in their research, it is imperative to revisit such problems. In this review three tests, T-maze spontaneous alternation task (T-CAT), Context dependent fear conditioning (CDFC), and Morris water maze (MWM) sensitive to hippocampal function, serve as illustrative examples for the potential problems. Spontaneous alternation tests are sometimes flawed because the handling procedure makes the test dependent on fear rather than exploratory behavior leading to altered alternation rates independent of hippocampal function. CDFC can provide misleading results because the context test, assumed to be a configural task dependent on the hippocampus, may have a significant elemental, i.e. cued, component. MWM may pose problems if its visible platform task is disproportionately easier for the subjects to solve than the hidden platform task, if the order of administration of visible and hidden platform tasks is not counterbalanced, or if inappropriate parameters are measured. Without attempting to be exhaustive, this review discusses such experimental problems and gives examples on how to avoid them.