We investigate the [CII] line intensity mapping (IM) signal from galaxies in the Epoch of Reionization (EoR) to assess its detectability, the possibility to constrain the L CII − SFR relation, and to recover the [CII] luminosity function (LF) from future experiments. By empirically assuming that logL CII = logA + γSFR ± σ L , we derive the [CII] LF from the observed UV LF, and the [CII] IM power spectrum. We study the shot-noise and the full power spectrum separately. Although, in general, the shot-noise component has a much higher signal-to-noise ratio than the clustering one, it cannot be used to put independent constraints on logA and γ. Full power spectrum measurements are crucial to break such degeneracy, and reconstruct the [CII] LF. In our fiducial survey S1 (inspired by CCAT-p/1000 hr) at z ∼ 6, the shot-noise (clustering) signal is detectable for 2 (1) of the 5 considered L CII −SFR relations. The shot-noise is generally dominated by galaxies with L CII 10 8−9 L (M UV ∼ −20 to −22), already at reach of ALMA pointed observations. However, given the small field of view of such telescope, an IM experiment would provide unique information on the bright-end of the LF. The detection depth of an IM experiment crucially depends on the (poorly constrained) L CII − SFR relation in the EoR. If the L CII − SFR relation varies in a wide logA -γ range, but still consistent with ALMA [CII] LF upper limits, even the signal from galaxies with L CII as faint as ∼ 10 7 L could be detectable. Finally, we consider the contamination by continuum foregrounds (cosmic infrared background, dust, CMB) and CO interloping lines, and derive the requirements on the residual contamination level to reliably extract the [CII] signal.