We present mid-infrared (MIR) imaging and spectroscopic data of the Seyfert 2 galaxy Mrk 1066 obtained with CanariCam (CC) on the 10.4 m Gran Telescopio CANARIAS (GTC). The galaxy was observed in imaging mode with an angular resolution of 0.24 ′′ (54 pc) in the Si-2 filter (8.7 µm). The image reveals a series of star-forming knots within the central ∼400 pc, after subtracting the dominant active galactic nucleus (AGN) component. We also subtracted this AGN unresolved component from the 8-13 µm spectra of the knots and the nucleus, and measured equivalent widths (EWs) of the 11.3 µm Polycyclic Aromatic Hydrocarbon (PAH) feature which are typical of pure starburst galaxies. This EW is larger in the nucleus than in the knots, confirming that, at least in the case of Mrk 1066, the AGN dilutes, rather than destroys, the molecules responsible for the 11.3 µm PAH emission. By comparing the nuclear GTC/CC spectrum with the Spitzer /IRS spectrum of the galaxy, we find that the AGN component that dominates the continuum emission at λ <15 µm on scales of ∼60 pc (90-100%) decreases to 35-50% when the emission of the central ∼830 pc is considered. On the other hand, the AGN contribution dominates the 15-25 µm emission (75%) on the scales probed by Spitzer /IRS. We reproduced the nuclear infrared emission of the galaxy with clumpy torus models, and derived a torus gas mass of 2×10 5 M ⊙ , contained in a clumpy structure of ∼2 pc radius and with a column density compatible with Mrk 1066 being a Compton-thick candidate, in agreement with X-ray observations. We find a good match between the MIR morphology of Mrk 1066 and the extended Paβ, Brγ and [O III]λ5007 emission. This coincidence implies that the 8.7 µm emission is probing star formation, dust in the narrow-line region, and the oval structure previously detected in the near-infrared. On the other hand, the Chandra soft X-ray morphology does not match any of the previous, contrary to what it is generally assumed for Seyfert galaxies. A thermal origin for the soft X-ray emission, rather than AGN photoionization, is suggested by the different data analyzed here.