Information hiding is an emerging research area which includes applications such as copyright management, tamper detection and content authentication. Existing chaos optimization and random position selection based watermarking embedding methods make claims for robustness and reversibility; however, reversibility becomes questionable when data capacity is high. In a published paper, authors proposed a watermarking algorithm based on a chaotic system, which claims to be reversible; however, based on our experimental results, the existing method suffers from overflow and underflow issues regardless of initial values.