We studied the disc of the unclassified B[e] star HD 50138, in order to explore its structure, and to find indications for the evolutionary status of this system, whether it is a young Herbig Be or a post-main-sequence star. Using high spatial resolution interferometric measurements from MIDI instrument (N-band) on the Very Large Telescope Interferometer, we analysed the disc size, the time-variability of the disc's thermal emission, and the spectral shape of the 10 µm silicate feature. By fitting simple disc models, we determined the inclination and the mid-infrared size of the disc, confirming earlier results based on a lower number of observations. We searched for mid-infrared temporal variability of different regions of the disc, and concluded that its morphology is not experiencing significant changes over the observed epochs. We characterized the mid-infrared silicate feature by determining the feature amplitude and the 11.3/9.8 µm flux ratio. The latter parameter is a good indicator of the grain size. The shape of the feature suggests the presence of crystalline silicate grains in the disc. The interferometric data revealed a strong radial trend in the mineralogy: while the disc's innermost region seems to be dominated by forsterite grains, at intermediate radii both forsterite and enstatite may be present. The outer disc may predominantly contain amorphous silicate particles. A comparison of the observed spectral shape with that of a sample of intermediate-mass stars (supergiants, Herbig Ae/Be stars, unclassified B[e] stars) implied that the evolutionary state of HD 50138 cannot be unambiguously decided from mid-IR spectroscopy. nandes et al. 2011), however, high spatial resolution observations can provide an efficient classification tool for them (Miroshnichenko 2007). A notable feature of unclassified B[e] stars is their high mass loss rates, compared to normal mainsequence stars, which can not be explained by the windtheory for main-sequence B-type stars (de la Fuente et al. 2015). Originally unclassified B[e] stars were thought to be isolated objects, but recently de la Fuente et al. (2015) reported that these objects can be also found in clusters. They estimated the age of the stars in the range from 3.5 to 6.5 Myr, which is out of the range for both pre-and post-main sequence origin, as the former phase lasts for ∼ 0.1 Myr, and the main-sequence lifetime is 15 − 20 Myr for a 12 M star.