Of all motor skills, manual reaching might be the one ability that matters most for infants’ perceptual, cognitive and social development. Reaching allows infants to learn about object properties, but also gives opportunities for socializing with others. The general aim of the present thesis was to study the importance of manual motor development in infancy from different perspectives; first, through examining stereopsis as a prerequisite for efficient reaching development, second, with regard to understanding others goal-directed reach actions by means of the mirror neuron system (MNS), and third, in relation to possible atypical development, with a specific focus on autism spectrum disorder (ASD). Study I shows that under monocular viewing conditions, infants at six, eight and 10 months of age perform slower and less accurate reaches. Longer times to object contact during monocular trials specifically imply that motor prediction is less effective when depth and distance information is compromised. Study II demonstrates that, by eight months of age, infants seem to have a MNS that functions in a similar manner to the adult MNS, thus activity can be registered over the motor cortex when infants simply observe an action they can master themselves. This activation is predictive, indicating anticipation of the goal of the observed reach. Study III indicates that infants at elevated familial risk for ASD present with reduced prospective motor control at 10 months of age. Compared to a low-risk control sample, high-risk infants perform reactive rather than predictive reach actions. Follow-up assessment at 36 months will show whether this measure can be used as a predictive diagnostic marker for ASD. The main contribution given by this work is the insight that it is important to take manual motor aspects into account when considering typical as well as atypical cognitive and social development, and in addition, that motor prediction is a key factor behind being able to timely execute and understand reaching actions.