Movement abnormalities are important symptoms in clinical neurology. Neurological diseases such as Parkinson's disease (PD), multiple sclerosis (MS), amyotrophic lateral sclerosis (ALS), stroke and pain present with abnormalities of motor function and severe locomotor deficits. Over the years, many methods have been developed to analyze locomotor behavior in small laboratory animals, particularly in rat. The CatWalk, a computer-assisted automated quantitative gait analysis system, allows rapid and objective quantification of a large number of gait parameters. In comparison to ink-test or similar approaches, previously used to assess static locomotor alterations, CatWalk can also obtain dynamic parameters. Despite the extensive use of the mouse in experimental models for neurological disease there have been relatively few studies on CatWalk in mouse. Furthermore, few information are available regarding methodological aspects on training prior to test. The aims of the present study are: (1) to provide reliable and reproducible results on over-ground locomotion in intact mouse using three different training protocol and, (2) to analyze the effect of training protocol on over-ground locomotion. In this study we tested three different experimental trainings, in order to get the best protocol for the use of the CatWalk in mice models. Animals were trained to run the walkway for different duration, either 11, 4 or 3 days, providing groups A, B and C respectively; mice of groups A and B were left to cross freely the walkway, while group C was forced to run only in one direction. Mice of groups A and B were subjected to a food deprivation (diet of 75% and then of 30%), while ad libitum food was available for mice of group C. Furthermore, mice of groups B and C were rewarded with high palatable food placed in a cage at the end of the runway, for motivating them. We objectively quantified a large number of gait parameters during over-ground locomotion and we focused for the first time on relation between animal weight and basic gait parameters because different training protocols can act as confounder for the interpretation of the results. The obtained results can thus serve as a template for studying changes in locomotor performance in mouse due to neurological disease or trauma to the CNS and may provide relevant information that can be advantageous to select a particular experimental protocol to be used in different models of movement disorders.