The functional and numerical responses of Habrobracon hebetor (Say) were assessed over 30 sequential generations (G) on different densities (2, 4, 8, 16, 32, 64, and 128 fifth-instar larvae) of the Mediterranean flour moth, Ephestia kuehniella (Zeller). Seven tested generations (G2_inf, G4_inf, G6_inf, G8_inf, G10_inf, G15_inf, and G30_inf) had already been naturally infected by a fungus species, whereas only the second generation (G2) had the colonies of both infected (G2_inf) and uninfected (G2_uninf) individuals. All infected generations, as well as the uninfected generation, showed a Type III functional response. A high variation was found in the handling times (T_h) of the parasitoid through 30 sequential generations, and G10_inf had the longest handling time. The shortest handling time and the maximum estimated attack rate (T/T_h) were observed in G2_inf. No significant difference in terms of the number of parasitized hosts was found among all infected generations, as well as between G2_inf and G2_uninf. Although the handling times in G2_inf and G2_uninf were close to each other, the attack coefficient of G2_uninf was more than that of G2_inf. The attack coefficient and handling time of infected generations increased from G2_inf to G10_inf and then decreased in the higher generations. The most plausible parameters of functional response of infected generations were observed in G_inf. A negative relationship between eggs laid and host densities was observed in the infected generations (G2_inf, G4_inf, G6_inf, G8_inf, and G10_inf). G2_uninf and, among infected generations, G_inf were the best generations for practical utilization in inundative release programs.