目前國內風力規範「建築物耐風設計規範及解說」相關條文大多引用美國ASCE 7-02(2002)之內容與參考日本AIJ(1996)之橫風向與扭轉向風力規定，而橫風向與扭轉向風力規定中的風力係數及風力頻譜計算式缺乏地況及高寬比之變數，使用上也非常保守。 本研究針對目前國內風力規範中橫風向與扭轉向風力設計所需的風力係數及風力頻譜進行探討，並與淡江大學風工程研究中心所做的試驗數據做比較且將其做非線性迴歸分析，目的在以簡化之公式擬合風力係數及風力頻譜，供未來在橫風向及扭轉向風力設計中使用。 本論文最後以風洞試驗數據為基準，選取斷面深寬比1/5、1/3、1/1、3/1及5/1與高寬比2、4及6之模型，將風洞試驗之頻譜及迴歸之頻譜結果以內政部建築研究所於101年度委託研究報告「設計風載重資料庫之應用研究」中所提到的風載重計算模式進行風載重計算並比較其計算結果，同時加入以國內風力規範計算之風載重，比較迴歸的成效。 規範計算之風載重在高寬比越高時結果越接近實驗值，在高寬比較低時則明顯低於實驗值，原因為規範的適用範圍及風載重計算模式不同所造成。比較迴歸計算之風載重與規範計算之風載重，迴歸計算之風載重幾乎較規範計算之風載接近實驗值。 At present, the majority of Taiwan wind code are cited from ASCE 7-02(2002), and the acrosswind and torsional wind loading regulations refers to the AIJ(1996). Therefore, the lack of variances of terrain and aspect ratio in the calculations of acrosswind and torsional wind coefficients and spectrums is also presented, resulted in a very conservative application of Taiwan’s wind code. This thesis focused on the investigation of wind coefficients and wind spectrums required for acrosswind and torsional wind loading analysis in Taiwan’s wind code. Comparisons with the data of tunnel experiments provided by Wind Engineering Research Center of Tamkang University were conducted as well. In addition, nonlinear regression analysis of the experimental data was performed. The purpose of this analysis is to fit simplified formula on wind coefficients and spectrums, which can be used for future acrosswind and torsional wind loading designs. Wind tunnel data including side ratio 1/5, 1/3, 1/1, 3/1 and 5/1, and aspect ratio 2, 4 and 6 were selected as the final comparison basis. Wind loadings calculated from wind spectrums of experiments and of nonlinear regression models based on the wind loading calculation model in the 2012 ABRI project:「Applications of Aerodynamic Database on Building Design Wind Loads」 were both studied and compared. In addition, the results of nonlinear regression were also check against wind loading calculated using Taiwan wind code. Wind loadings calculated using Taiwan wind code for high aspect ratio buildings are close to experimental data, but are obviously smaller than experimental results for the low aspect ratio. The reasons are caused by the applicable range of the code and different wind loading calculation models. Wind loadings calculated from nonlinear regression are always closer to experimental results than using wind code.