Mercuric reductase which originated from a recombinant Escherichia coli PWS1 was puri®ed and immobilized on a chemically modi®ed diatomaceous earth support. The mercury reduction kinetics, pH dependence, storage stability, and reusability of the immobilized enzyme were investigated. Four dyes were examined for their electron transfer ef®ciency with the soluble and bound mercuric reductase. Continuous mercury detoxi®cation by the immobilized mercuric reductase was also performed in ®xed-bed processes. The effects of bed-length, mercury loading rate, and electron donor on the performance of the ®xed beds were assessed. Immobilized mercuric reductase exhibited substrate-inhibition-type kinetics with a maximal activity (1.2 nmol Hg mg À1 protein s À1 ) occurring at an initial Hg 2 concentration of 50 mmol dm À3 . The optimal pH was 7.0 for the soluble and immobilized mercuric reductase, but the immobilized enzyme maintained higher relative activity for less favorable pH values. Immobilization of the enzyme appeared to signi®cantly enhance its storage stability and reusability. Of four arti®cial electron donors tested, azure A (5 mmol dm À3 ) demonstrated the highest relative activity (78%) for soluble mercuric reductase. For the immobilized enzyme, neutral red (5 mmol dm À3 ) gave a relative activity of nearly 82%. With a ®xed-bed, the mercury-reducing ef®ciency of using neutral red was only 30±40% of that obtained using NADPH. Fixed-bed operations also showed that increased bed length facilitated mercury reduction rate, and the optimal performance of the beds was achieved at a ¯ow rate of approximately 100±200 cm 3 h À1 .