Although a population bias toward right-hand preference is observed at the early stage of grasping, hand preference fluctuates in infancy. Given these fluctuations, one can wonder whether testing a young infant on a single occasion gives reliable results of its handedness. Very few studies have evaluated short-term test-retest reliability. This was the goal of this study in which 21 infants aged 9-15 months were tested for handedness every day for a total of 5 sessions. The infants were given a classical handedness baby test. Their handedness index (HI) and their category of handedness were compared across sessions. The results show that at the group level the distribution of handedness does not differ significantly across the five sessions. At the individual level, only 19% of infants were categorized as right-handed at all five sessions while 52.4% were consistent in using more one hand than the other across the five sessions (right hand: 47.6%). Most of the fluctuations across sessions occurred between being lateralized and non-lateralized rather than between being right-handed and left-handed. These figures indicate that testing handedness at that age gives fairly reliable results in terms of direction of hand preference, but less so in terms of degree.