The exponential growth in transmission of multimedia over the Internet and unsecured channels of communications is putting pressure on scientists and engineers to develop effective and efficient security schemes. In this paper, an image encryption scheme is proposed to help solve such a problem. The proposed scheme is implemented over three stages. The first stage makes use of Rule 30 cellular automata to generate the first encryption key. The second stage utilizes a well-tested S-box, whose design involves a transformation, modular inverses, and permutation. Finally, the third stage employs a solution of the Lorenz system to generate the second encryption key. The aggregate effect of this 3-stage process insures the application of Shannon’s confusion and diffusion properties of a cryptographic system and enhances the security and robustness of the resulting encrypted images. Specifically, the use of the PRNG bitstreams from both of the cellular automata and the Lorenz system, as keys, combined with the S-box, results in the needed non-linearity and complexity inherent in well-encrypted images, which is sufficient to frustrate attackers. Performance evaluation is carried out with statistical and sensitivity analyses, to check for and demonstrate the security and robustness of the proposed scheme. On testing the resulting encrypted Lena image, the proposed scheme results in an MSE value of 8923.03, a PSNR value of 8.625 dB, an information entropy of 7.999, NPCR value of 99.627, and UACI value of 33.46. The proposed scheme is shown to encrypt images at an average rate of 0.61 Mbps. A comparative study with counterpart image encryption schemes from the literature is also presented to showcase the superior performance of the proposed scheme.