除了美國外，經濟較為發達的國家近年來總生育率絕大多數都在2.1 的人口替代水準之下，台灣地區在民國93年的總生育率首次降至1.2以下的歷史新低，民國94年的嬰兒出生數進一步降至約20萬人，預計又將創新紀錄，加速台灣地區的人口老化。由於台灣地區的生育率變化較大，筆者之前的研究發現其他各國的生育率模型不見得適用於台灣，本文希望能以系統式探討各國生育率，確定哪些生育率模型適用經濟較為發達的國家。本文引用台灣、日本（亞洲）、荷蘭（歐洲）、美國（美洲）等經濟發達國家的實證資料，加上電腦模擬與敏感度分析，評估包括Gamma 、Lee-Carter、主成份分析（Principal Component Analysis）、單一年齡組個別估計法、擴散模型（Diffusion Model）等較為常用的模型，比較這些國家配適結果的異同。 實證分析發現台灣、日本、美國都是以年齡組生育率個別配適擴散模型的總生育率最佳，荷蘭則是單一年齡組個別估計法；在年齡組生育率的預測方面，台灣、日本、荷蘭都是以年齡組生育率個別配適擴散模型最好，美國則是以單一年齡組個別估計最好。電腦模擬的結果則以單一年齡組個別估計法最佳，個別配適擴散模型次之。

Except in the U.S., the total fertility rates of developed countries have been smaller than the replacement level of 2.1. Without exception, the total fertility rates in Taiwan dropped to a historic low (1.2) in 2004 and are expected to be even smaller in 2006. In this study, we used the fertility data of four developed countries, Taiwan, Japan, the Netherlands and the USA (representing countries in Asia, Europe, and America), to fit the frequently used models (Gamma, Lee-Carter, Principal Component Analysis, Age-Group Fertility Rate, and Diffusion models) and determine which model has the best fit. We found that, for the total fertility rates, the best model in Taiwan, Japan, and the USA is our modified diffusion model, while in the Netherlands it is the age-group fertility rate model. For the age-specific fertility rates, the best model in Taiwan, Japan, and the Netherlands is the modified diffusion model, while in the USA it is the age-group fertility rate model. In addition, we use computer simulation to check the relative stability and relative efficiency of these fertility models. The age-group fertility rate model and the modified diffusion model also outperform other models. In conclusion, we recommend the readers use these two models.