Protein proteinase inhibitors are widely distributed in plants, animals and microorganisms. They can be conveniently grouped since most frequently they inhibit proteinases belonging only to a single mechanistic class. Protein inhibitors of serine proteinases have been most extensively studied. They are strictly competitive inhibitors forming 1:1 complexes with the enzymes they inhibit. In these complexes, all activities of the enzyme are completely abolished. The inhibitors are substrates for the enzyme they inhibit at a unique peptide bond called the reactive site peptide bond (one for each inhibitory domain). However, compared to normal substrates where the enzyme-substrate and enzyme-product complexes dissociate very readily here, the complexes are very stable. Serine proteinase inhibitors can be divided into at least 13 families. Within each family the position of the reactive site and the closure of disulfide bridges can be inferred by homology. In enzyme-inhibitor complexes, about 10-15 residues of the inhibitor are in contact with the enzyme. Their specific nature strongly affects both the strength and the specificity of enzyme-inhibitor interaction. In all cases where the sequences of many inhibitors from the same family can be compared, the contact residues are not strongly conserved--instead, they are hypervariable. This raises major problems but also offers huge opportunities to those concerned with the role of inhibitors in biology and in medicine.