The performance of neural codes to represent attributes of sensory signals has been evaluated in the vertebrate peripheral and central nervous system. Here, we determine how information signaled by populations of neurons is modified by plasticity. Suprathreshold neuronal responses from a large number of neurons were recorded in the juvenile mouse barrel cortex using dithered random-access scanning. Pairing of one input with another resulted in a long-lasting, input-specific modification of the cortical responses. Mutual information analysis indicated that cortical plasticity efficiently changed information signaled by populations of neurons. The contribution of neural correlations to the change in mutual information was negative. The largest factor limiting fidelity of mutual information after pairing was a low reliability of the modified cortical responses.