土地開發分析法是開發人員與估價師評估土地價格的重要方法之一，然而不同從業人員針對相同標的估值往往不盡相同，是估價人員的偏誤造成價格上的差異？抑或是估價人員運用土地開發分析法時，各項投入參數差異所產生？本研究以台北市公開標售成交地價與上市建設公司評估台北市地價分別分析，利用@Risk尋找土地開發分析法各項投入參數可能的機率分配型態，再以蒙地卡羅模擬分析建立土地開發分析價格的機率模型，並利用平均絕對百分比誤差（MAPE）、命中率（hit Rate）及交插驗證(cross-validation)做為機率模型預測估值優劣程度的衡量標準，比較「公開標售」組與「評估地價」組兩者機率模型何者預測能力。結果發現「評估資料」組的估值準確性較佳，「公開標售」組所建立之機率模型較不具預測效果。而三種適合度檢定之機率模型則是以A-D適合度檢定所建立之機率模型表現最佳(MAPE為14.98%，誤差±10% Hit-rate為38.46%，誤差±20% Hit-rate為71.43%)。本文透過機率模型的建立，考量不動產開發隱含之風險，呈現估值的結果由定值轉變為區間；考量不同估價人員或開發業者間，對於相同投入參數不同認知的情形；本模型將對地價敏感性較大之投入參數由定值轉為機率密度函數，較估價人員或開發業者主觀認定各項投入參數之變動更為客觀；可應用於大量估價，亦可作為估價人員或開發業者評估土地時，價格之參考 / Land development analysis is one of the developers and valuers assessment of land price method, different practitioners, however, is often not the same for the valuation of the same subject, valuers bias caused by the difference in price? Or valuers use of the land development analysis, the various input parameters the differences? In this study, turnover in Taipei sale by public tender premium and public listed construction company to assess the land value of Taipei City were analyzed, Looking for land development analysis the probability distribution of possible input parameters using the @ Risk, useing Monte Carlo simulation analysis to establish the land development analysis of the price of the probability model, and using the mean absolute percentage error (MAPE), the hit rate and cross-validation as a probability model to predict the measure of the quality and extent of the valuation, "open auction” group and “assess the premium” group between the two probability model may be predictive ability. The results showed that the valuation accuracy of the assessment information group better probability model created by the sale by public tender group compared with no predictive effect. Three kinds of fit test of the probability model is based on the AD for degree verification to establish the probability model the best performing (MAPE of 14.98%, the error is ± 10% Hit-the rate is 38.46%, the error is ± 20% the Hit-the rate is 71.43% ). Through the establishment of the probability model, consider the real estate development implied the risk of showing the results of the valuation by the valuation change interval; considerations between the different valuation or development industry, for the same input parameters of different cognitive situations; this model will have premium The larger sensitivity of the input parameters from the given value to the probability density function over valuation or development industry subjective judgments more objective changes of the input parameters; can be applied to a large number of valuation, assessment of land may be used as the valuation or development industryWhen the reference price.