A novel approximation technique for reduction of higher order continuous systems utilizing sine cosine algorithm (SCA) is proposed. The parameters of reduced-order model (ROM) of the higher order continuous system (HOCS) are obtained by minimizing the normalized errors in between the Markov parameters and time moments of ROM and HOCS. The main advantage of the proposed technique is that it always engenders stable ROM for stable HOCS. The stability of ROM is ascertained utilizing Hurwitz criterion. The SCA is able to explore different regions of search space, evade local optima. This exploits promising regions of a search space during optimization efficaciously. The efficacy of the proposed technique is illustrated with the avail of one test system. A comparative study is additionally accomplished for results obtained utilizing SCA and other optimization techniques. The analysis denotes that SCA outperforms when compared to other algorithms.