A range-wide sampling of Picea rubens Sarg. populations with a representative sampling of Picea mariana (Mill.) B.S.P. was made to investigate the variation within P. rubens and the nature and extent of hybridization between the two species. Factor analysis using 14, 21, and 24 characters, successively, indicated that the clusters of the parent species were quite discrete. Variation was found to be continuous within but not between the species, and gene flow was not more in one direction than the other. Species were found to maintain their centroides with a very low level of hybridization, and introgressive hybridization was found to be of a very limited nature.Controlled crosses indicated that the crossability of the parent species with the distantly allopatric Picea omorika ranged from 36 to 71%, while the crossability of the sympatric P. rubens × P. mariana and the reciprocal was from 1 to 3%. Other interspecific crosses were considered. Variation in the P. rubens × P. mariana F1 was substantial and sufficient to explain most of the variation observed in the occasional hybrid swarm found in nature. Backcrossing data indicate that it is not more successful than parent species crosses. Regeneration data from a soil site moisture catena for both species were evaluated to estimate the number of hybrids moving into the population relative to the parent species. These were found to be very low. A homeostatic process involving an array of both genetic and ecological barriers is proposed to explain the persistence of P. rubens in the presence of P. mariana and other selection pressures. A comparison is made of the relationships between P. glauca and P. engelmannii and that of P. rubens and P. mariana. It is concluded that the latter are not a species pair in the same sense as the former.