In China, Cultivated Land Productivity Evaluation (CLPE) is carrying out country-wide on various scales, e.g. country, province, municipality, city and county, with the《Technical Rules for Productivity Survey and Quality Assessment of Cultivated Lands》 (TRPSQACL) as guidance. In order to minimize negative effects of the modeling based on the subjective evaluation index system, discussion is done in this paper on analytic logic of CLPE, so as to build up a “thought bridge” linking professional knowledge to the physical model of cultivated land productivity. CLPE is just a kind of land productivity evaluation (LPE) and the prevailing LPE methods could be used as reference in forming scientific analytic logic for building a CLPE index system. LPE methods can roughly be sorted into 3 categories. The first one is the category of qualitative classification and grading, such as Land Capability Classification of America (LCC), Soil Fertility Capability Classification of America (FCC), Chinese Land Resource Evaluation Map at 1:100,000 scale. This category of methods is used to classify and grade land resources according to the type and extent of obstacles in land use on all spatial scales, but it does not have a consistent index system. The second is one of quantitative scoring methods, such as Storie Index Soil Rating of America, Soil Quality Rating of Germany (SQR), etc., characterized by explicit index systems and scoring rules, and fast computation with the aid of GIS, but they are only applicable to evaluation on small scales for the sake of accuracy. And the third is one consisting of methods combining classification with scoring, such as Agro Ecological Zone of FAO, Land Potential Productivity suggested by MOSS and Agricultural Land Quality Grading of China. They are characterized by dividing the region to be evaluated into several agricultural zones, for which appropriate evaluation indices are selected zone-specifically, and then calculating potential of phototemperature or climate contributing to land productivity on the large spatial scale as the first step, and modifying the score with soil conditions and site-specific environmental conditions on the small scales as the second step. So they are applicable to evaluation on all spatial scales. Different from the above-listed 3 categories of methods, TRPSQACL specifies a scoring method for CLPE without considering the limitation of this method being unapplicable to evaluation on large scales. So, it has brought about a series of problems in the application, e.g. absence of specific task orientation suitable to all spatio-temporal scales in evaluation, neglect of scale suitability in selecting indices, frequent failure to have the principles of dominance and independence embodied. By utilizing the viewpoints of geoscience, agronomy and system science, this paper has built analytic logic to solve the above-mentioned problems. The paper holds: 1) for CLPE, the implications of task, technical routes and indices to evaluation all possess the feature of spatio-temporal scale. 2) Evaluation indices should be divided into two groups, of which one may directly indicate satisfaction degree of the crop with light, temperature, water and nutrients while the other may indirectly do. It is advisable to choose indices with stability and indirect evaluation indices for CLPE on large scales, which is characterized by long-term effectiveness and low spatial resolution and focus on exposing geographical obstacle and resource constraints in utilizing cultivated land in a region For CLPE on small scales aiming at real-time status of cropland productivity with high spatial resolution, indices featuring instantaneity and directness are recommended. 3) Land-crop-management should be viewed as a holistic system and significances of the indices to cropland productivity should be dialectically analyzed since not only natural and management land conditions but also the match degree of land conditions and crop requirements vary with land system. And 4) the natural and artificial land conditions that jointly indicate nature of a certain aspect, e.g. light, temperature, water and nutrients, should be viewed as a whole, an integrated index, that should not be dissevered in the evaluation index system.