Digital-to-Analog Converter (DAC) is a crucial building block limiting the accuracy and speed of many signal processing and telecommunication systems. To achieve high speed and high resolution, the current-steering architecture is almost exclusively used. Three important issues for current-steering DAC design are addressed in this dissertation. In a current-steering DAC design, it is essential that a designer determine the minimum required current source accuracy to overcome random current mismatch and achieve high linearity with guaranteed yield. Simple formulas are derived that clearly exhibit the relationship between the standard deviation of unit current sources, the bits of resolution, the INL/DNL, and the soft yield of DAC arrays. It is shown that these formulas are very effective for optimizing the DAC segmentation so as to achieve high performance and high yield with minimal area and power consumption. To overcome random mismatch effects without any trimming, the current source array of a high-accuracy DAC is usually rather large, causing the gradient errors in these arrays to become significant. How gradient errors affect the DAC linearity and how to compensate for them through switching sequence optimization is analyzed in the second part of this dissertation. To overcome technology barriers, relax the requirements on layout and reduce the sensitivities of DACs to process, temperature and aging, calibration is emerging as an attractive solution for the next-generation high-performance DACs, especially as process feature size keeps shrinking and supply voltage is reduced correspondingly. A new foreground calibration technique suitable for low-voltage environment is presented in the third part of this dissertation. It can effectively compensate for current source mismatches, and achieve high linearity with small die size and low power consumption. The dynamic performance of the DAC is also improved due to the dramatic reduction of parasitic effects. To demonstrate this technique, a 14-bit prototype was designed and fabricated in a 0.13u digital CMOS process. It is the first 14-bit CMOS DAC ever reported that operates with a single 1.5V power supply, occupies an active area less than 0.1mm2, and requires only 16.7mW at 100MHz sampling rate, but still maintains state-of-art linearity and speed.