Fixed heterosis resulting from favorable interactions between the genes on their homoeologous genomes in an allopolyploid is considered analogous to classical heterosis accruing from interactions between homologous chromosomes in heterozygous plants of a diploid species. It has been hypothesized that fixed heterosis may be one of the causes of low classical heterosis in allopolyploids. We used Indian mustard (Brassica juncea, 2n = 36; AABB) as a model system to analyze this hypothesis due to ease of its resynthesis from its diploid progenitors, B. rapa (2n = 20; AA) and B. nigra (2n = 16; BB). Both forms of heterosis were investigated in terms of ploidy level, gene action and genetic diversity. To facilitate this, eleven B. juncea genotypes were resynthesized by hybridizing ten near inbred lines of B. rapa and nine of B. nigra. Three half diallel combinations involving resynthesized B. juncea (11×11) and the corresponding progenitor genotypes of B. rapa (10×10) and B. nigra (9×9) were evaluated. Genetic diversity was estimated based on DNA polymorphism generated by SSR primers. Heterosis and genetic diversity in parental diploid species appeared not to predict heterosis and genetic diversity at alloploid level. There was also no association between combining ability, genetic diversity and heterosis across ploidy. Though a large proportion (0.47) of combinations showed positive values, the average fixed heterosis was low for seed yield but high for biomass yield. The genetic diversity was a significant contributor to fixed heterosis for biomass yield, due possibly to adaptive advantage it may confer on de novo alloploids during evolution. Good general/specific combiners at diploid level did not necessarily produce good general/specific combiners at amphiploid level. It was also concluded that polyploidy impacts classical heterosis indirectly due to the negative association between fixed heterosis and classical heterosis.