The crystal structure of benzoylacetone (1-phenyl-1,3-butanedione, C(10)H(10)O(2); P2(1)/c, Z = 4) has been determined at 300, 160 (both Mo Kalpha X-ray diffraction, XRD), 20 (lambda = 1.012 Å neutron diffraction, ND) and 8 K (Ag Kalpha XRD), to which should be added earlier structure determinations at 300 (Mo Kalpha XRD and ND, lambda = 0.983 Å) and 143 K (Mo Kalpha XRD). Cell dimensions have been measured over the temperature range 8-300 K; a first- or second-order phase change does not occur within this range. The atomic displacement parameters have been analyzed using the thermal motion analysis program THMA11. The most marked change in the molecular structure is in the disposition of the methyl group, which has a librational amplitude of approximately 20 degrees at 20 K and is rotationally disordered at 300 K. The lengths of the two C-O bonds in the cis-enol ring do not differ significantly, nor do those of the two C-C bonds, nor do these lengths change between 8 and 300 K. An ND difference synthesis (20 K) shows a single enol hydrogen trough (rather than two half H atoms), approximately centered between the O atoms; analogous results were obtained by XRD (8 K). It is inferred that the enol hydrogen is in a broad, flat-bottomed single-minimum potential well between the O atoms, with a libration amplitude of approximately 0.30 Å at 8 K. These results suggest that at 8 K the cis-enol ring in benzoylacetone has quasi-aromatic character, in agreement with the results of high-level ab initio calculations made for benzoylacetone [Schiøtt et al. (1998). J. Am. Chem. Soc. 120, 12117-12124]. Application [in a related paper by Madsen et al. (1998). J. Am. Chem. Soc. 120, 10040-10045] of multipolar analysis and topological methods to the charge density obtained from the combined lowest temperature X-ray and neutron data provides evidence for an intramolecular hydrogen bond with partly electrostatic and partly covalent character, and large p-delocalization in the cis-enol ring. This is in good agreement with what is expected from the observed bond lengths. Analysis of the total available (through the Cambridge Structural Database, CSD) population of cis-enol ring geometries confirms earlier reports of correlation between the degree of bond localization in the pairs of C-C and C-O bonds, but does not show the dependence of bond localization on d(O.O) that was reported earlier for a more restricted sample. It is suggested that the only reliable method of determining whether the enol hydrogen is found in a single or double potential well is by low-temperature X-ray or (preferably) neutron diffraction.