Crude bacterial collagenase was chromatographed on DEAE‐cellulose to yield three peaks with proteolytic activity: an arginine esterase (DEAE‐1), gelatinase (DEAE‐2) and a caseinolytic activity (DEAE‐3). The arginine esterase and gelatinase activity fractions were slightly contaminated with each other but neither possessed caseinolytic activity; the caseinolytic fraction was devoid of arginine esterase and gelatinase activities. In addition, crude collagenase was fractionated by ZnII‐affinity chromatography to produce a gelatinase peak (ZnII peak 1), which was free from arginine esterase and caseinolytic activities. The four fractions were compared to crude collagenase in their ability to liberate rat hepatocytes by using either liver slices or a standard perfusion technique. Compared to crude collagenase (0.05–0.1% w/v), which produced 70–80% liver digestion with approximately 80% cell viability, digestion with equivalent quantities of the isolated enzymic activities was relatively poor. Gelatinase activity (ZnII peak 1) was wholly ineffective and DEAE‐1 and DEAE‐2 each possessed only slight digestive properties. Hepatocyte liberation by the caseinolytic activity, DEAE‐3, was partially successful (30–40% digestion, 25–30% viability) but only a portion of liver tissue was digested regardless of the quantity of DEAE‐3 used. However, by mixing certain fractions before perfusion two gelatinase‐dependent, cell‐releasing mechanisms were identified: (a) DEAE‐3 with ZnII peak 1 and (b) DEAE‐1 mixed with either DEAE‐2 or ZnII peak 1. Each system compared creditably with the digestive properties of an equivalent activity of crude collagenase. At present we are attempting to determine any differences between hepatocytes produced by the two enzymic mechanisms.