We analyzed molecular radiation in the spectra of the very bright Benešov bolide. The Benešov bolide appeared over the Czech Republic on May 7, 1991 and reached an absolute magnitude of –19.5. It was caused by a meteoroid larger than 1m. Small meteorites of various mineralogical types were recovered recently (Spurný et al. [2014] Astron. Astrophys., 570, A39, 14 p.). The spectrum of the bolide, recorded on two photographic plates, is probably the richest meteor spectrum ever obtained. It contains hundreds of atomic emission lines, continuous radiation and molecular bands, and covers the whole bolide trajectory from the altitude of 90–20km. In this paper we focus on identification and analysis of molecular bands. The identification of FeO, CaO, AlO, and MgO, reported earlier (Borovička and Spurný [1996] Icarus, 121, 484–510) was confirmed. In addition, radiation of N2 was probably detected. The oxides were best seen in the wake and in the radiating cloud left at the position of the bolide flare at the altitude of 24.5km. Trace of N2 was seen only in the meteor at lower altitudes. FeO bands are present in the spectra from the highest altitudes. We suppose that FeO was ablated directly in molecular form at high altitudes. CaO was first detected just below 50km and its intensity, relatively to FeO, strongly increased toward lower altitudes. AlO, which is similarly refractive as CaO, behaved as FeO rather than CaO at lower altitudes. MgO was observed only in the radiating cloud. The spectrum of the cloud is unique because it contains almost no atomic lines. We compared the data with theoretical calculations of the presence of molecules in the mixture of meteoric vapors and air at various altitudes and temperatures. CN and TiO were not found. The upper limit of CN is in agreement with theory for ordinary chondrite meteoroid. Most of carbon should be in fact present in the form of CO, but CO bands are too weak to be detected. The non-detection of TiO can be explained by the fact that temperature in the wake and the cloud was lower than needed for the presence of TiO bands. However, AlO was found to be about 40 times more abundant than MgO, although comparable abundances are expected. The explanation may be that the abundances are in fact comparable but there are non-equilibrium conditions in the radiating cloud with the excitation temperature of MgO lower than that of AlO. The difference may be caused by higher ablation temperature of Al. Another non-equilibrium effect is the observed difference between the rotational (∼1000K) and vibrational (∼3000K) temperature of AlO molecules. This can be explained by short hydrodynamic timescale and the fact that vibrational relaxation time is significantly longer than rotational relaxation time. The vibrational temperature therefore could not decrease so quickly during the cooling and expansion of the cloud because of insufficient number of collisions. FeO and CaO could not be analyzed in detail, because their molecular constants, especially transition probabilities, are not well known. The increase of the CaO/FeO ratio with decreasing altitude could be, nevertheless, explained in scope of equilibrium chemistry.