Low yield is one of the main challenges of rice production in Ghana due to poor production practice, environmental stresses and plant genotype. The choice of an efficient breeding procedure depends to a large extent on knowledge of the genetic system controlling the character to be selected. This experiment was done to study the inheritance pattern of aroma attribute of rice and estimate the heritability for some yield attributes in rice (Oryza sativa L.). Six generations viz., P1, P2, F1, F2, BCP1 and BCP2 of a cross between IET6279 and IR70445-146-3-3 were used for the study. A randomized complete block design (RCBD) was used with three replications. All the F1 and BCP1 plants of the cross were non-aromatic indicting that the gene controlling aroma in the donor parent was recessive. The segregation ratio of aromatic to non-aromatic plants was 1:3 in F2 and 1:1 in BCP2 plants, confirming the monogenic inheritance of aroma. This shows that there would be a high probability of success in selecting for aroma using pedigree breeding in early generations of F2. High broad sense heritability estimates were observed for all the characters studied except for number of fertile spikelets per plant, suggesting that the influence of the environment was minimal for these characters. Hence, these traits would respond to selection owing to their high genetic variability and transmissibility. Thus, selection based on phenotypic value of these characters would be reliable and effective with low broad sense heritability for number of fertile spikelets per plant, indicating that influence of the environment was high for these characters. Therefore, superior genotypes selection based on phenotypic performance for this trait may not be effective. The analysis of gene effects revealed that additive, dominance and epistatic effects were involved in the inheritance of all the traits studied. However, additive effect, in general was higher in magnitude than dominance gene effect in all the characters. All the traits manifested non-allelic interactions, indicating that epistasis is determined to some extent by the genotypes used for the study. Thus, in the presence of epistasis, grain width showed significant duplicate type of gene interaction which restricted the scope of simple selection for the characters studied. Therefore delaying selections to later generations using recurrent selection will enhance success in improving yield attributes in all genotypes studied. Key words: Aroma, inheritance, additive, gene action, heritability, monogenic, non-additive, recessive.