To elucidate pathways from specific genes to complex behaviors, assays of mouse behavior need to be valid, reliable and replicable across laboratories. Behavioral assays are proving to be as complex as the intricate cellular and molecular pathways that are the main interest of many mouse users. There is no perfect behavioral test, but we propose some aphorisms to stimulate discussion that is necessary for continued progress in task development. For maximal utility, a behavioral test should yield valid data for most of the commonly used inbred mouse strains. Tests of simple, ubiquitous behaviors usually yield meaningful data for most mice, especially when based on automated scoring or on simple physical measures that are likely to be replicable across laboratories. Extreme test scores resulting from non-performance on a task can inflate the apparent reliability of a test, and devious adaptations to a task can undermine its validity. The optimal apparatus configuration for certain genetic or pharmacological analyses might depend on the particular laboratory environment. Despite our best efforts, the mice will continue to win some innings.