The spatial coding in the hippocampus (HPC) and its emergence during learning is well studied, but it is still unclear how the spatial information in the medial prefrontal cortex(mPFC) is changing between different learning stages. We recorded multiple unit activity using a combination of 24 movable tetrodes for the HPC and a Neuropixels silicon probe with 384 channels for the mPFC. We monitored the formation of spatial representations in both regions while the rat performed a spatial memory task on the radial eight-arm maze. The animal started in the center of the 8-arm maze and needed to locate three rewards. After the completion of the task, the animal needed to return to the center for a 2-minute pause/rest. We recorded 8 trials per day and subsequent 2 hours of sleep. The animals required at least three days to reach close to optimal behavioral performance to locate all three goals in the initial configuration. The data from 404 cells in HPC and 1602 cells in mPFC was recorded from three animals with three days each. Using the spatial representation in HPC and mPFC separately, we were able to predict animal position, arm, and direction. Furthermore, we detected replayed trajectories in both regions during awake and sleep periods. These data demonstrate the involvement of mPFC in spatial coding during gradual spatial learning in mazes.