I have developed a model for optic flow selectivity that is based on physiological and anatomical features of small-field local motion-sensitive cells (bushy T cells) and their outputs to wide-field collator neurons (lobula plate tangential cells). I have used this model to identify two features that are crucial for generating optic flow selectivity: the broadness of the spatial patterns of synaptic connections (innervation matrices) from motion detectors to collators, and the relative contributions of excitatory and inhibitory connections. The results show that optic flow selective properties of the model are quite robust. Although both small-field physiological properties of the motion detectors and wide-field connection patterns can be optimized, there is a broad range of circuit configurations within which significant departures from “optimal” parameters are of little consequence for the performance of the network. This means that substantial variability in the evolutionary, developmental or homeostatic processes that shape optic flow processing circuits is possible without endangering their most important computational properties. A general implication is that sensory systems across the animal kingdom may have evolved robust mechanisms that do not rely on precise network parameters. For additional information, see Douglass and Strausfeld (2000a; 2000b).