A model of scale-free neuronal networks, which consists of heterogeneous Fitzhugh-Nagumo neurons and time-delayed coupling, was constructed. Then, we explored the nontrivial effects of heterogeneity and time-delayed coupling on the resonance dynamics by numerical simulation in this model. When the delays in the coupling are absent, the result has shown that the response of the neuronal networks to an external subthreshold periodic signal is optimized at an intermediate heterogeneity, namely, an appropriate tuned level of heterogeneity can induce resonance in the neuronal networks. This phenomenon was also confirmed to be robust to the changes of the coupling strength. Most importantly, we find that the delays in the coupling have significant influences on the resonance dynamics. It is revealed that proper delays can induce multiple resonances in the neuronal networks, which appears at each multiple of the oscillation period of the signal. Moreover, the performance of fine tuned delays in inducing multiple resonances can also be clearly observed when the heterogeneity is within an appropriate range.