The ability of the brain to extract meaning from sensory stimuli depends on the computational capabilities of precisely connected neuronal networks in the cerebral cortex. Despite variations between cortical areas and species, neurons receiving similar inputs or sharing similar response properties often aggregate into fine-scale functional networks. In the mature visual cortex (V1), for example, neurons are not randomly connected, but form specific local and long-range circuits, which may influence the neuronal selectivity for different features of visual scenes, such as their orientation, direction and position. How such fine-scale functional circuits emerge during postnatal development from an immature neuronal network remains a fundamental question in neuroscience. We are addressing this question in mouse V1 by directly relating patterns of excitatory synaptic connectivity to visual response properties of neighbouring layer 2/3 pyramidal neurons at different postnatal ages, using two-photon calcium imaging in vivo in anaesthetized mice and multiple whole-cell recordings in vitro. Our previous work revealed that pyramidal neurons connect preferentially if they respond similarly to visual stimuli. At eye-opening, although neural responses were already highly selective for visual stimuli, neurons responding to similar visual features were not yet preferentially connected, indicating that the emergence of feature selectivity does not depend on the precise arrangement of local synaptic connections. After eye opening, local connectivity reorganised extensively, as more connections formed selectively between neurons with similar visual responses, and connections were eliminated between visually unresponsive neurons, while the overall connectivity rate did not change. We propose a unified model of cortical microcircuit development based on activity-dependent mechanisms of plasticity: neurons first acquire feature preference by selecting feedforward inputs before the onset of sensory experience, after which patterned input drives the formation of functional subnetworks through a redistribution of recurrent synaptic connections.