Microorganisms preferentially allocate their resources to synthesis of phosphorus- and nitrogen-acquiring enzymes when phosphorus and nitrogen availability is low, respectively, according to the resource allocation model for ecoenzyme synthesis. However, validity of this model for sulfur (S)-acquiring enzymes across various soils remains uncertain. Here we showed that the resource allocation model for ecoenzyme synthesis is valid for the S-acquiring enzyme, arylsulfatase, across arable (Acrisols, Andosols, and Fluvisols) and forest (Andosols and Cambisols) soils. A significant negative relationship of the ratio of arylsulfatase to β-d-glucosidase activities was observed with soluble and adsorbed sulfate concentration that was stronger than the relationships with other indices of S availability (autoclave S, organic S, total S, and mineralizable S), indicating that microorganisms produced more arylsulfatase when the concentration of this S form was low in soils. The ratio of arylsulfatase to β-d-glucosidase activities in forest soils (mean, 1.55) was significantly higher than that in arable soils (mean, 0.50), whereas the opposite trend was observed for soluble and adsorbed sulfate concentration (mean, 33 and 127 mg S kg−1, respectively). Because soluble and adsorbed sulfate is known to be plant-available, the ecoenzymatic stoichiometry could also be useful for evaluating soil S plant-availability.