The use of active narrow-band interference (NBI) suppression in direct-sequence spread-spectrum (DS-SS) communications has been extensively studied. In this paper, we address the problem of optimum coding-spreading tradeoff for NBI suppression. With maximum likelihood decoding, we first derive upper bounds on the error probability of coded systems in the presence of a special class of NBI, namely, multi-tone interference with orthogonal signatures. By employing the well-developed bounding techniques, we show there is no advantage in spreading, and hence a low-rate full coding approach is always preferred. Then, we propose a practical low-rate turbo-Hadamard coding approach, in which the NBI suppression is naturally achieved through iterative decoding. The proposed turbo-Hadamard coding approach employs a kind of coded spread-spectrum signalling with time-varying spreading sequences, which is sharply compared with the code-aided DS-SS approach. With a spreading sequence of length 32 and a fixed bandwidth allocated for both approaches, it is shown through extensive simulations that the proposed turbo-Hadamard coding approach outperforms the code-aided DS-SS approach for three types of NBI, even when its transmission information rate is about 5 times higher than that of the code-aided DS-SS approach. The use of the proposed turbo-Hadmard coding approach in multipath fading channels is also discussed.