Comparisons of mitochondrial DNA (mtDNA) sequences may establish maternal phylogenies, date species divergence, provide an analysis of hybrid zones, and generate conclusions about evolutionary and population dynamics (Clark and Lyckegaard, 1988; MacRae and Anderson, 1988; McClean and Hanson, 1986; Ovendon and White, 1990). Most models dealing with mtDNA diversity assume neutrality of alternate alleles (Arnold et al., 1988; Birky et al., 1983; Chapman et al., 1982; DeSalle et al., 1987; Kimura, 1983; Kimura and Ohta, 1969; Neigel and Avise, 1986). If the assumption of neutrality is not met, evolutionary conclusions based on neutrality may be compromised (Lansman et al., 1983; Ovendon and White, 1990). Several lines of evidence challenge the validity of the neutral theory. Clark and Lyckegaard (1988) make the point that phylogenies established from nuclear and mitochondrial clocks should be identical if the neutral theory is true, but such equality is not the rule. In reviewing the evidence for and against the neutral theory of mtDNA, Clark (1984) concluded that variation of mtDNA sequences is not random, which suggests evolutionary constraints on sequence variation. The present research demonstrates a difference in fitness between cybrid and normal lines of tobacco (Nicotiana tabacum) that differ only in the mitochondrial genome. A cybrid is a plant with a single parental nucleus and a hybrid cytoplasm produced by mixing the two parental cytoplasms (Belliard et al., 1979). The two lines have a nearly isogenic nuclear background (N. tabacum) but the cybrid has a recombinant mitochondrial genome, the result of mixing mitochondria from two allopatric tobacco species (Belliard et al., 1979). These plants allow direct comparison of functional differences that can be attributed only to the mitochondria or to mitochondrial-nuclear interactions. Results indicate that mitochondrial alleles are not neutral in tobacco derived from two different species. This is relevant to the behavior of alternative mitochondrial genomes in natural populations, particularly in hybrid zones where introgression of foreign mtDNA would be expected (Ferris et al., 1 983; Powell, 1983; Saunders et al., 1986; Tegelstrom, 1987).