Exhaustive-substitution studies, where many amino acid replacements are individually tested at all positions in a natural protein, have proven to be very valuable in probing the relationship between sequence and function. The broad picture that has emerged from studies of this sort is one of functional tolerance of substitution. We have applied this approach to barnase, a 110-residue bacterial ribonuclease. Because the selection system used to score barnase mutants as active or inactive detects activity down to a level that can be approached by nonenzyme catalysts, mutants that test inactive are essentially devoid of enzymatic function. Of the 109 barnase positions subjected to substitution, only 15 (14%) are vulnerable to this extreme level of inactivation, and only 2 could not be substituted without such inactivation. A total of 33 substitutions (amounting to 5% of the explored substitutions) were found to render barnase wholly inactive. The profoundly disruptive effects of all of these inactivating substitutions appear to result from either (1) replacement of a side chain that is directly involved in substrate binding or catalysis, (2) replacement of a substantially buried side chain, (3) introduction of a proline residue, or (4) replacement of a glycine residue. Although substitutions of these types are functionally tolerated more often than not, the system used here indicates that only these sorts of substitution are capable of single-handedly reducing catalytic function to, or nearly to, levels that can be achieved by nonenzyme catalysts.