Work on enzymatic cross-linking of globular food proteins has mainly focused on food functional effects such as improvements of gelation and enhanced stabilization of emulsions and foams, and on the detailed biochemical characterization of the cross-linking chemistry. What is still lacking is a physical characterization of cluster formation and gelation, as has been done for example, for cluster formation and gelation during heat-induced protein aggregation. Here we present preliminary results along these lines. We propose that enzymatic cross-linking of apo-alpha-lactalbumin is a good model system for studying the problem of cluster formation and gelation during enzymatic cross-linking of globular proteins. We present initial results on cluster sizes produced when crosslinking dilute solutions of apo-alpha-lactalbumin with a range of cross-linking enzymes: microbial transglutaminase, horseradish peroxidase, and mushroom tyrosinase. These results are used to highlight similarities and differences between different enzymes, when acting on the same substrate. Next we consider cluster growth and gelation in somewhat more detail for the specific case of cross-linking by horseradish peroxidase, under the periodic addition of H2O2. Upon increasing the initial concentration of apo-alpha-lactalbumin, at a fixed enzyme-to-substrate ratio and fixed reaction time, the size of the clusters at the end of the reaction increases rapidly, and above a critical concentration, gelation occurs. For the conditions that we have used, gelation occurred at very low initial apo-alpha-lactalbumin concentrations of 34% (w/v), indicating a very dilute cross-linked protein network, with a low average number of cross-links per protein. It is found that reactive protein monomers are first rapidly (1-2 h) incorporated into small covalent clusters. This is followed by a much slower phase (up to about 12 h) in which the small clusters are coupled together to form much larger covalent protein clusters. Consistent with this two-step mechanism, atomic force microscopy shows that the covalent protein clusters are very heterogeneous and seem to consist of smaller subclusters.