The establishment of certain patterns of mystacial vibrissae in mice has been the aim of an extensive breeding program carried on in this laboratory since 1977. In a companion paper we have reported on variations in this pattern in an outbred population of ICR mice. Starting with 21 ICR animals we bred, mostly by brother-sister mating, for 13 bilaterally symmetric patterns of mystacial vibrissae characterized by the presence (or absence) of supernumerary whiskers (SWs). The strains are classified as follows: I, a mouse strain with the standard pattern; II, eight strains bred for the occurrence of SWs at a given site or sites; and III, four mouse strains bred for a maximal number of SWs in different regions of the whiskerpad. Commonly, SWs occur in regions that coincide with the zones of mergence between the three facial processes except for two class II strains in which we bred for SWs in the "straddler" row of vibrissae, and for one class III strain, in which we cultivated the tendency (that appeared late in our program) to have SWs at the crest of a facial process. For classes I and II we analyzed the results for about 18 generations in terms of "improvement," meaning an increase in the percentages of animals with the desired phenotype together with a decreased frequency of undesired SWs. For class III, success in breeding meant the increase of the mean number of the desired SWs. All results led to the same conclusion: there is a genetic basis for the occurrence of SWs. The side preference of a particular SW is not strain dependent. It disappears in those class I and II strains in which almost 100% of animals obtained the desired phenotype. The increase in number of SWs in one zone of mergence does not depend on the presence of SWs in the other. Where tested, we almost always found a representation of an SW in a topologically equivalent location within the "barrelfield" area of the somatosensory cerebral cortex. Except for some diseases early in the breeding program, and some side effects of inbreeding that were eliminated, the population was without obvious defects. Where tested, there was no correlation between the occurrence of SWs and sex. The observed variations in pattern of mystacial vibrissae and their genetic background led us to propose a morphogenetic model for the formation of the pattern of mystacial vibrissae.