Four enzyme preparations containing β-glucuronidase, of bacterial, mammalian, and molluscan origin, have been shown to be equally effective in liberating 17-ketosteroids (17-KS) of the 5β-(etiocholane) configuration in normal urine. The bacterial preparation releases steroids of the 5α-(androstane) configuration more rapidly than do the molluscan enzymes and with much greater ease than does the liver enzyme. In view of the data obtained it seems unlikely that the striking difference between the bacterial and liver enzymes can be due to the hydrolysis of some labile conjugate, such as sulphate, by the former and not by the latter. Possibilities that the difference is due to the hydrolysis of an unknown type of urinary conjugate by the bacterial preparation, or to the low specificity of the bacterial β-glucuronidase, are discussed. The high degree of hydrolysis of 17-KS conjugates by the bacterial enzyme followed by solvolysis suggests this as a most useful hydrolytic procedure.