Economic barriers preventing commercialization of lignocellulose-to-ethanol bioconversion processes include the high cost of hydrolytic enzymes. One strategy for cost reduction is to improve the specific activities of cellulases by genetic engineering. However, screening for improved activity typically uses "ideal" cellulosic substrates, and results are not necessarily applicable to more realistic substrates such as pretreated hardwoods and softwoods. For lignocellulosic substrates, nonproductive binding and inactivation of enzymes by the lignin component appear to be important factors limiting catalytic efficiency. A better understanding of these factors could allow engineering of cellulases with improved activity based on reduced enzyme-lignin interaction ("weak lignin-binding cellulases"). To prove this concept, we have shown that naturally occurring cellulases with similar catalytic activity on a model cellulosic substrate can differ significantly in their affinities for lignin. Moreover, although cellulose-binding domains (CBDs) are hydrophobic and probably participate in lignin binding, we show that cellulases lacking CBDs also have a high affinity for lignin, indicating the presence of lignin-binding sites on the catalytic domain.