This article provides parameter optimization processes of sinusoidal signal generators (SSGs) based on a digital-to-analog converter (DAC) for electrical impedance spectroscopy (EIS) applications. The SSG, which is the most power-hungry building block in EIS systems, generates a sinusoidal signal by using a DAC. To achieve high accuracy for the overall EIS system, high linearity is required for the sinusoidal signal. Thus, the SSG's DAC is typically operated with a high oversampling ratio (OSR) and a large number of quantization levels (L-DAC) at the expense of increased power consumption, large area, and high complexity. For efficient use of the power and area in the SSG, it is necessary to optimize the OSR, L-DAC, and the order of the low-pass filter (LPF) (N-LPF). In this article, optimal design parameters of SSGs, which can achieve highly accurate EIS systems with low complexity, are presented. First, the minimum OSR and NLPF for lowering the magnitude error to less than 0.1% are presented. Then, optimal quantization levels of finite-resolution DACs are found for sufficient accuracy and harmonic tones. In addition, the accuracy and harmonics of odd-number OSR cases are analyzed and compared with even-number OSR cases. According to the results, it is possible to design an SSG that only differs from the ideal sinusoidal signal by approximately 0.1% by using OSR <= 32, N-LPF <= 2, and L-DAC <= 256.