Abstract ATP-dependent deoxyribonuclease was first purified up to 580-fold from a crude extract of Bacillus subtilis W168 by ammonium sulfate fractionation, agarose and DEAE-cellulose column fractionation. Enzyme instability was reduced by use of Tris-HCl media and an in-line dialysis apparatus directly connected to a chromatographic column. During the course of the enzyme purification, an inhibitor to this enzyme was detected in the crude extract prepared by sonication and was partially purified by ammonium sulfate fractionation and DEAE-cellulose column chromatography. Based on various criteria, it was concluded that the inhibitor is fragmented DNA or DNA in complex with some other compounds. The ATP-dependent DNase of B. subtilis has relatively narrow limits of ATP concentration for its optimal activity, and there is a marked inhibition above the optimal ATP concentration (20 µm). However, a rather constant level of residual activity is found above 100 µm, up to at least 1 mm ATP. The enzyme also requires Mg2+ for its activity; Km for this is 1.02 mm. The ATP-dependent DNase of B. subtilis has unusal substrate specificities for native and denatured DNA. The enzyme (Fraction VI) requires ATP to hydrolyze native DNA, but is not at all necessary for the hydrolysis of denatured DNA. The activity on the native DNA with optimal concentration of ATP (20 µm) was 3-fold greater than on the denatured DNA with or without ATP, while at a higher ATP concentration (500 µm), the factor was 0.5-fold. Among chemical inhibitors tested, β,γ-methyleneadenosine triphosphate inhibited the enzyme activity with native DNA as a substrate, while with denatured DNA the activity remained unaltered. The possibility of a contaminating enzyme which would hydrolyze single strand DNA preferentially appears unlikely because of the coincidence of the two activities on both agarose and DEAE-cellulose column chromatographies, together with the salt inactivation behavior. A glycerol gradient centrifugation of the above partially purified enzyme raised the purity to approximately 4800-fold. This procedure, however, made a drastic change in the property of the enzyme. Thus, the ratio of activities on native DNA to denatured DNA at 20 µm ATP increased from 3 in Fraction VI to more than 30. A possibility of removing a contaminating enzyme specific to denatured DNA as the substrate was made unlikely. DNA-dependent adenosine triphosphatase activity was found inseparable from the purified enzyme.