對高度科技專業性領域之行政處分之司法審查密度，我國專業法律鮮少明定司法控制密度，相關問題從而成為法律解釋適用問題。我國實務與學說向來繼受德國判斷餘地理論，以功能法的觀點出發，行政法院只有低度審查權，只有當系爭決定構成「判斷逾越」或「判斷濫用」時，始構成撤銷事由。但德國自一九九○年考試評分案等系列裁判以來，經由學界的詮釋，已經略有改變，除了功能法之外，也應補充以系爭決定對於當事人之基本權干預的嚴重性，此外，也應斟酌系爭案型與規定有無其他之促成合理決定之機制。功能法之觀點、基本權干預之嚴重性與系爭法律所提供之擔保行政決定合理性之機制多寡，乃構成行政法院控制密度高低的衡量基礎。 我國學界與大法官解釋並未引進前述新的見解，從而對人權的保障不週，在不得已的情形下，台北高等行政法院以事涉人民生命健康法益之嚴重侵害為理由，而巧妙地提高「判斷濫用」的標準，誠有見地。 行政法院的控制密度高低，也發生於高度科技性法律條文的詮釋（多以行政規則為之）。德國法提出形式面之授權要求、由行政機關聽取專業領域專家意見而以自我責任作出決定，以及實質面之以無恣意方式完成、廣泛納入相關議題之科學技術上之專業意見、相關內容不得過時矛盾或得被學術動搖、相關行政規則的作成程序須能擔保正確性。 至於重大開發案之判斷或衡量之疏漏，我國行政程序法採補正原則，對人權保護未必周延，但反面言之，動輒撤銷全案是否有助於全球化競爭下之我國招商，也非無討論餘地。

The judicial review of administrative adjudication in cases of great public interests, so the case of siting of the Science Park of Mid-Taiwan, is a problem to be reconsidered in Taiwan. As for judicial review of questions of law, the judicial branch in charge has apparently not the specialized knowledge and capability of the final determination about the factfinding, self restraints is necessary, but in case of great public interests, the German model has changed since 1990, a new standards of judicial review is born, it requires the balancing of functional boundary of courts, intensity of the intervention of human rights and all other relevant factors of the case-type is necessary. As for judicial review of questions of law, so it must be taken into account, that judges are real specialist of questions of law, and therefore competent to overrule the legal opinions proposed by the administrative agency, and therefore annul the administrative adjudication. But however, about the question of fact, the stander of review could be somehow lowered, according to the structures and regulations of the policy field in concern. And yet, the German model remains probably an exception of the world. In the siting case, the Taipei Administrative Court, by way of ”abuse of judgement”, took an intensive review of questions of fact and annulled the adjudication, which reflects somehow the changing idea expressed in the new Germen model.