In this work, 4 yeasts (Pichia wickerhammi, Candida wickerhammi, Pichia stipitis CBS5776, Pichia stipitis CBS5876), and 5 fungi (Theivalia spp., Thermoascus aurantiacus, Tyromyces palustris A, Aspergillus niger, Trichoderma harzianum), known to excrete enzymes capable of hydrolyzing polysaccharides found in association with $\beta$-D-mannans, were assessed for their ability to degrade $\beta$-D-mannans. The $\beta$-D-mannanase activity found in the culture filtrate of T. harzianum was selected for further study and a 'cellulase-free' $\beta$-D-mannanase isolated from culture filtrates of T. harzianum grown on medium supplemented with 1% w/v locust bean gum studied. $\beta$-D-Mannanase activity was detected in T. harzianum culture filtrates from media supplemented with 1% (w/v) Avicel, locust bean gum galactomannan, konjac root glucomannan, or spruce wood water-solubles. Medium supplemented with 1% (w/v) mannose did not induce $\beta$-D-glucomannanase or $\beta$-D-galactomannanase activity. However, when 0.5% (w/v) $\beta$-D-galactomannan was added with mannose, $\beta$-D-mannanase activity was detected in the culture filtrate. Growth of the fungus on mannan-rich locust bean gum resulted in the highest specific $\beta$-D-glucomannanase and $\beta$-D-galactomannanase values. A zymogram assay was developed to selectively detect $\beta$-D-mannanase activity in crude culture filtrates. The presence of different polysaccharides in the growth medium resulted in different $\beta$-D-mannanase zymogram profiles. Analyses of the protein profiles of the culture filtrates separated by isoelectric focusing revealed several bands having $\beta$-D-mannanase and endoglucanase activity. A protein band having $\beta$-D-mannanase activity but lacking detectable cellulase activity was identified. This enzyme was purified to homogeneity via a sequence involving ultrafiltration, ion exchange and gel filtration. This 'cellulase-free' $\beta$-D-mannanase has the highest reported pI for a fungal $\beta$-D-mannanase. The isolated enzyme had a molecular weight of 42.9 $\pm$ 4 kD, an optimum temperature of 60-65$\sp\circ$C, an optimum pH of 5.8, a pI of 6.55, and possessed at least 75% of maximum activity over a pH range from 3.21-6.8. Enzymatic activity was stable during 12 months of storage at 4$\sp\circ$C. $\beta$-D-mannanase activity was resistant to pepsin, $\alpha$-chymotrypsin, trypsin, and Staphylococcus V8 protease. The effect of metal ions, detergent and solvents on $\beta$-D-mannanase activity was also determined. Although the enzyme did not degrade Avicel or Solka floc, it did associate with both celluloses. Enzyme associated with these celluloses remained active towards locust bean gum galactomannan. The overall efficiency of the enzyme (V$\sb{\rm max}$/K$\sb{\rm m})$ for the target substrate, locust bean gum galactomannan, was reduced by the presence of 1.0% w/v Avicel. Of the four $\beta$-D-mannans tested, the enzyme had greatest overall efficiency towards konjac glucomannan. However, deacetylation of konjac glucomannan lowered the efficiency of the enzyme by 41.5%. (Abstract shortened by UMI.)