Estimation of steel consumption is an essential part of the implementation of quota design. Increasing accuracy of this estimation will not only help control the design phase, but also back up project decision and make it possible to conduct procurement plan earlier, which will shorten project life cycle. This paper adopts Support Vector Machine method among the numerous methods to apply it to this estimation in high-rise residential buildings and achieve relatively high accuracy through case verification. Furthermore, the influences of input variables' variation on the outcome are revealed via sensitivity analysis, which can sustain project decision.